FINAL REPORT OF THE 2004 ELECTION DAY SURVEY

Transcription

1 FINAL REPORT OF THE 2004 ELECTION DAY SURVEY Submitted to the U.S. Election Assistance Commission Kimball W. Brace, Principal Investigator Dr. Michael P. McDonald, Consultant EAC Survey Analysis Support (EAC 0524) September 27, 2005

2 CONTENTS EXECUTIVE SUMMARY PART 1: INTRODUCTION Survey Design and Development... 2 Development Timeline... 2 Survey Questions... 2 Applicability... 4 Survey Response... 4 Election Jurisdictions... 4 Coverage... 5 Data Tabulation... 8 Data Measurement... 9 Table Format... 9 Cross Tabulations Type of Voting Equipment Changed Voting Equipment since Statewide Voter Registration Database Election Day Registration Provisional Ballot Acceptance No Excuse Absentee Balloting Early Voting Section 203 Language Minority Requirements Section 5 Preclearance of Voting Procedures Regions Urban to Rural...14 Size of Jurisdiction (VAP) Race and Ethnicity Median Income High School Education Battleground States Presidential Margin of Victory Red versus Blue Jurisdictions... 17

3 2004 Election Day Survey Report Executive Summary, Page 2 September 27, 2005 PART 2: ELECTION DAY SURVEY RESULTS Chapter 1: Population Estimates Applicability and Coverage Population Estimates Analysis of Estimates Chapter 2: Voter Registration Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 3: Ballots Counted Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 4: Turnout Source Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 5: Absentee Ballots Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 6: Provisional Ballots Applicability and Coverage Historical Context Survey Results Analysis of Survey Results

4 2004 Election Day Survey Report Executive Summary, Page 3 September 27, 2005 Chapter 7: Drop-Off Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 8: Overvotes and Undervotes Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 9: Voting Equipment Usage Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 10: Voting Machines Applicability and Coverage Survey Results Analysis of Survey Results Chapter 11: Voting Equipment Malfunctions Applicability and Coverage Survey Results Chapter 12: Poll Workers Applicability and Coverage Historical Context Survey Results Analysis of Survey Results Chapter 13: Polling Places Applicability and Coverage Historical Context Survey Results Analysis of Survey Results

5 2004 Election Day Survey Report Executive Summary, Page 4 September 27, 2005 Chapter 14: Disability Applicability and Coverage Historical Context Survey Results Analysis of Survey Results PART 3: RECOMMENDATIONS ON FUTURE DATA COLLECTION Appendix A. Survey Instrument Appendix B. Election Glossary

6 2004 Election Day Survey Report Executive Summary, Page 5 September 27, 2005 List of Figures and Supplemental Tables Table 2a. State Voter Registration Deadlines Table 2b. Voter Registration Statistics, Table 2c. State Rankings for Registration Calculations Table 3a. Electoral Drop-Off Rates, Table 3b. Turnout Rates Sorted Table 6a. Provisional Ballot Usage, Sorted Table 9a. Voting Equipment for the November 2004 General Election Table 9b. Voting Equipment Manufacturers, November 2004 General Election Table 9c. Voting Equipment Usage, 2004 General Election Table 11a. Reported Voting Equipment Malfunctions by Equipment Type Table 11b. Reported Reasons for Electronic Voting Equipment Malfunctions Table 11c. Reported Reasons for Optical Scan Voting Equipment Malfunctions Table 13a. Number of Precincts Nationwide, Figure 4.1 Ballots Source, Figure 7.1 Drop-Off Rate, Figure 9.1 Voting Equipment , Percent of Counties Figure 9.2 Voting Equipment , Percent of Registered Voters Figure 9.3 Voting Equipment Usage, 2004: Jurisdictions Figure 9.4 Voting Equipment Usage, 2004: Registration Figure 11.1 Jurisdictions with Equipment Malfunctions

7 EXECUTIVE SUMMARY For the first time in this nation s over 230-year history, the federal government has taken an assessment of the election process used across the land. It has not been perfect, it contains holes and mistakes errors that might even be such that conclusions point in the wrong direction, but it is a start. As a philosopher once said, A mile s journey begins with the first step. Let the journey begin. This is a report by Election Data Services, Inc. of the Election Day Survey for November 2, 2004, administered by the United States Election Assistance Commission (EAC) as mandated by the 2002 Help America Vote Act of The Election Day Survey represents the largest and most comprehensive survey of voting and election administration practices ever conducted by a U.S. government organization. Questions on the Election Day Survey covered voting statistics on voter registration; total ballots cast by mode of voting; specific statistics on absentee and provisional voting; votes for federal offices; the number overvotes and undervotes cast for each federal office; and the number of precinct, polling places, and poll workers. Questions covered election administration of voting equipment, reported equipment failures, disability accesses to polling places, and sufficient number of poll workers. State election administrators in the 50 states, the District of Columbia, and four territories Guam, Puerto Rico, American Samoa, and the U.S. Virgin Islands were requested to provide responses to the Election Day Survey. State administrators, in turn, went to the local level (county or township) and sought the data from the local election administrators. All state election administrators except American Samoa and Guam participated in the Election Day Survey. Election Data Services was contracted to compile the responses to the Election Day Survey and provide analysis. The principal investigator was Kimball W. Brace, president of Election Data Services. The statistical consultant was Dr. Michael P. McDonald, assistant professor of government and politics in the Department of Public and International Affairs at George Mason University. Support personnel included research analysts, database programmers, and administrative assistants, all employees of Election Data Services. Administering the Election Day Survey The Election Day Survey requested voting and election administration information from all states and responses were received from 6,568 local election administration jurisdictions in the United States and four of its territories. Forty-three questions were asked of each of these local election administration jurisdictions. In total, 282,424 responses to individual question items were received. This was the first time the Election Day Survey was administered, and as is typical with baseline surveys, many issues were identified in administering the survey. These issues included: State and local election administrators do not share common terminology for survey items, for example, of what constitutes an absentee ballot or a poll

8 2004 Election Day Survey Report Executive Summary, Page 2 September 27, 2005 worker. Interpretation of the survey items by election administrators resulted in uneven reporting, sometimes within the same state. Some state election administrators altered the method of response an electronic spreadsheet and sent the spreadsheet to local election administrators to complete. Some survey questions were altered by central election administrators, which caused further uneven reporting among states. Some states transmitted local administrators responses without recompiling the responses into the format requested by the EAC, requiring additional effort by Election Data Services to compile Election Day Survey responses. Election Data Services identified many data-entry errors. In some cases, particularly those that were major outliers, we identified the error, asked the primary reporting source for clarification, and made a correction. However, we did not have the resources to validate all quarter-of-a-million-plus individual items on the survey and we are certain that many errors remain. Election administrators were not given enough lead time to anticipate the compilation of statistics for their responses to the Election Day Survey, which resulted in some information being lost before it could be collected. Despite the problems in administering the Election Day Survey, we believe that reliable information was obtained for many of the questions and that our analysis illuminates some of the successes and challenges to election administration in the United States. However, we caution that our findings are only valid to reporting jurisdictions we cannot make any inferences for unreported jurisdictions and that the reliability of some responses reduces the overall validity of our analysis. Furthermore, we strongly recommend that consumers of the raw data accompanying this report be cautious when identifying problematic cases. Too often, we found a simple data-entry error or a unique method of reporting data were responsible for the outlier cases that we observed. We make three main recommendations to the EAC to improve future data collection efforts: Election Data Services recommends that the EAC hold two symposiums, first, of state election administrators, and, second, of a larger pool of consumers of election data in order to produce accurate and consistent definitions of election administration statistics. A set of common definitions will increase the reliability of future data collection. We recommend that the symposiums be held in the near future to allow election administrators time to implement any new procedures that they may need to conform to the new definitions. Election Data Services recommends that the next Election Day Survey be conducted by a method that provides interactive quality assurance checks. Such a system might be Web-based or through a spreadsheet. Validating responses at the time of data entry will greatly reduce the number of data-entry errors.

9 2004 Election Day Survey Report Executive Summary, Page 3 September 27, 2005 Election Data Services recommends that the EAC expand its clearinghouse role to include the ongoing funding and collection of precinct-based registration, turnout, election returns, precinct maps, polling place information, sample ballots, manuals, and other election administration items to assist in the analysis of the voting process in this country. Further, and more detailed, recommendations are included in the recommendation chapter of this study and readers are encouraged to review that section. General Findings We categorize and discuss the Election Day Survey items in 12 areas: 1. Voter registration: counts of active and inactive voter registration 2. Total ballots counted: overall statistics on total ballots cast 3. Turnout Source: total ballots counted by mode of voting, cast in polling place, absentee, early, and provisional 4. Absentee ballots: number of absentee ballots requested, returned, counted, and not counted 5. Provisional ballots: ballots cast and ballots counted 6. Drop-off: the difference between total ballots counted and the ballots with a vote recorded for a federal office, for president, Senate, and U.S. House 7. Overvotes and undervotes: the number of overvotes (a ballot with more than one recorded vote for a candidate for an office) and the number of undervotes (a ballot with no recorded vote for a candidate for an office, for president, Senate, and U.S. House) 8. Voting equipment usage by election jurisdictions 9. Voting machine statistics: within election jurisdictions, the total number and number per precinct and polling place, and location of counting of ballots within polling place or central counting 10. Poll workers: number of poll workers and the number of jurisdictions reporting inadequate number of poll workers 11. Polling places: relationship between polling places and precincts 12. Disability: access by wheelchair, visually impaired, and physically disabled Voter Registration On voter registration, 6,512 of the 6,568 jurisdictions reported a total voter registration of 177,265,030. Of these, 4,879 jurisdictions reported active voter registration of 165,877,539, and 3,049 jurisdictions reported inactive registration of 21,695,013. Active and inactive voter registration does not sum to total registration because half of the states include and the other half exclude inactive registration in their total registration statistics.

10 2004 Election Day Survey Report Executive Summary, Page 4 September 27, 2005 For jurisdictions responding to the survey, the reported total voter registration constituted 79.5 percent of the voting age population (all persons age 18 and older residing in the United States; VAP) or 86.0 percent of citizen voting age population (CVAP). Among jurisdictions reporting total voter registration, those with higher education, higher income, Election Day registration, more rural and small-town in nature, and those found in battleground states tended to have higher rates of registration. Among jurisdictions reporting inactive voter registration, those that had the largest populations, jurisdictions with lower education, those covered by Section 203 of the Voting Rights Act, predominantly Hispanic jurisdictions, and those jurisdictions without statewide voter registration databases all had higher rates of inactive voters than other jurisdictions. Total Ballots Counted For ballots counted, 6,488 of 6,568 jurisdictions reported total ballots counted of 121,862,353. For reporting jurisdictions, this constituted 70.4 percent of voter registration, 55.8 percent of VAP, and 60.7 percent of the CVAP. However, a number of states and localities, almost one-sixth of the nation, still only report the votes cast for the highest office on the ballot as their total turnout, not the actual number of persons who participated in the election. Total number of ballots cast as a percentage of CVAP was higher in jurisdictions with higher education, higher income, predominantly non-hispanic White jurisdictions, those suburban in nature, jurisdictions not covered by Section 5 or Section 203 of the Voting Rights Act, those with Election Day registration, those without early voting, and those jurisdictions in battleground states. Turnout Source Of the jurisdictions that reported the mode of voting, 73.5 percent of total ballots cast were reported to have been cast in a polling place on Election Day, 13.3 percent were reported to have been cast as an absentee ballot, 23.5 percent were reported to have been cast as an early vote, 1.2 percent were reported to have been cast as a provisional ballot, and the remaining 23.1 percent were cast in an unknown manner. We note that Kansas and Texas report all votes cast prior to Election Day as an early vote. Washington reported that 68.7 percent of its vote was absentee, while Texas reported that 47.7 percent of its vote was cast early. There was some duplication of counting, especially where absentee ballots are counted on Election Day. Absentee Ballots Of those reporting jurisdictions, 16,870,660 absentee ballots were requested, 14,851,332 were returned (88.7 percent), and nearly all, or 14,740,215 (96.9 percent), were counted. Generally, those jurisdictions with fewer absentee ballots requested as a percentage of voter registration had a higher rate of return. This was true in jurisdictions with no excuse absentee voting and those permitting early voting. We speculate that the lower return rate is related to more difficult request criteria that deter less committed absentee voters from requesting a ballot. Jurisdictions with statewide voter registration databases reported similar request rates but a higher return rate, suggesting that better administration of registration roles improved the processing of absentee ballots.

11 2004 Election Day Survey Report Executive Summary, Page 5 September 27, 2005 Higher levels of absentee balloting and lower rates of return were found in higher education and income jurisdictions and in larger population, urban areas. We also noted this pattern in Section 203 covered jurisdictions, but we found that predominantly Hispanic and predominantly non-hispanic Native American jurisdictions reported both lower rates of absentee ballot request and absentee ballot return. The most common reason for absentee ballots to be rejected was that there was no voter signature. Provisional Ballots For those reporting jurisdictions, 1,901,591 provisional ballots were reported to have been cast. Of those, 1,225,915 were reported counted (or 64.5 percent). Provisional ballots were used by 2.56 percent of the persons casting ballots on Election Day, which also amounted to 1.25 percent of all registered voters for the election. The states of Washington and Alaska had the highest rates of provisional ballots cast, both over 10 percent of the ballots cast in the precincts. Jurisdictions that allowed provisional ballots to be cast jurisdiction-wide had higher rates of provisional ballots cast, as a percentage of total voter registration, and provisional ballots counted. Jurisdictions with a statewide voter registration database reported fewer provisional ballots cast, though a similar rate of acceptance. Section 203 covered jurisdictions reported a much higher rate of casting provisional ballots. Over one million provisional ballots, more than half of all provisional ballots cast, were cast in these jurisdictions even though they constituted an eighth of jurisdictions reporting provisional ballots. The much higher rate of casting provisional ballots was not offset by the higher rate of counting provisional ballots in these jurisdictions. Related, predominantly Hispanic jurisdictions reported the highest rate of casting provisional ballots, followed by predominantly non-hispanic Native American jurisdictions. Higher incidences of casting provisional ballots were found in urban and high population density areas, but these jurisdictions also had higher rates of counting provisional ballots. Rates of counting provisional ballots also tended to increase with the income and education level within a jurisdiction. The most common reason for rejecting provisional ballots was that the voter was not registered. Drop-Off For reporting jurisdictions, drop-off for the presidential election was reported at 1,160,985 or 1.02 percent of ballots cast for president. Drop-off for the Senate elections was reported at 5,676,784 or 6.86 percent of ballots cast for Senate. Drop-off for the U.S. House races was reported at 11,669,373 or percent of the ballots cast for U.S. House. Drop-off is most associated with competition. With a high degree of competition, drop-off in the 2004 presidential election was the lowest in a post-world War II presidential election. Among the Senate elections, drop-off was lowest in the most highly contested elections. Among the U.S. House elections, direct measures of competition were not asked on the Election Day Survey, however, elections in Section 5 and Section 203 covered jurisdictions reported higher rates of drop-off, which is consistent with these jurisdictions tending to be heavily Democratic and have relatively uncontested U.S. House elections. Among types of voting equipment, paper and punch card jurisdictions report about twice the presidential drop-off than other jurisdictions, while lever machines had the lowest drop-off rate.

12 2004 Election Day Survey Report Executive Summary, Page 6 September 27, 2005 Higher levels of drop-off in presidential, Senate, and U.S. House elections were related to lower levels of education and income. Overvotes and Undervotes For reporting jurisdictions, in the presidential election 133,289 overvotes or 0.23 percent of total ballots cast for president were reported; 863,872 undervotes or 0.91 percent of total presidential ballots cast for president were reported. In the Senate elections, 49,100 overvotes or 0.11 percent of total ballots cast for Senate were reported; 2,488,016 undervotes or 3.80 percent of total ballots cast for Senate were reported. In the U.S. House elections, 56,173 overvotes or 0.12 percent of total ballots cast for U.S. House were reported; 5,077,325 undervotes or 6.27 percent of total ballots cast for U.S. House were reported. The overall pattern of overvotes and undervotes underscores conventional wisdom that overvotes are a true error by voters while the majority of undervotes tend to be true abstention from the election in question. The percentage of overvotes across federal elections is small and within a similar range, while undervotes tend to increase with less competition in the election and the lack of importance of the office in the minds of most voters. Still, we find variation in overvotes and undervotes. Perhaps of most interest is rates of overvoting and undervoting in relation to the type of voting equipment. Jurisdictions using punch card and paper voting systems reported the highest overvotes as a percentage of total ballots cast. Jurisdictions using electronic systems reported a low percentage of undervotes, but jurisdictions using lever systems also reported a low rate, as did jurisdictions using multiple systems. Optical scan jurisdictions tended to fall in the middle. Jurisdictions with the lowest income and education levels tended to report the highest percentage of overvotes and undervotes. Rural and small voting age population size jurisdictions tended to report the highest percentage of overvotes, usually followed by urban or the largest population jurisdictions. Predominantly Hispanic jurisdictions tended to report the highest percentage of overvotes, and a generally high percentage of undervotes for all offices. Voting Equipment Usage About 14 percent of the jurisdictions failed to report what kind of voting equipment was in use for the 2004 general election. For those that did, more than one quarter of the nation s election jurisdictions used paper ballots, but because of their small size, only 1.8 percent of the registered voters voted in this manner. Nearly 40 percent of the nation s registered voters used optical scan systems in 2004, and 25 percent used electronic systems. Another 12 percent utilized lever machines and nine percent were still voting with punch cards. Thirteen states and territories had uniform voting equipment in use. Optical scan systems are used statewide in Alaska, Arizona, Oklahoma, Oregon, and Rhode Island. Electronic systems cover the states of Delaware, Georgia, Maryland, and Nevada, while lever machines are uniformly used in New York. The District of Columbia and Hawaii have mixed systems in place jurisdiction-wide. Voting Machines Only 20 states provided information on the number of actual machines in use for voting. Only onethird of the punch card jurisdictions provided any data on number of units, and much of that data

13 2004 Election Day Survey Report Executive Summary, Page 7 September 27, 2005 was probably incorrectly reported anyway. Fewer than half of the optical scan, lever, and electronic system jurisdictions reported the number of units used. With those limitations in mind, we have calculated that electronic system jurisdictions average three devices per precinct and slightly over five devices per polling place. The highest ratio of machines per location occurred in the state of Maryland, where between nine and 10 devices were used. Due to the high cost of lever machines, they average only 1.5 machines per precinct and slightly over three machines per polling place. The maximum number per precinct and polling place occurred in North Carolina and Wyoming, with slightly more than three per precinct and six per polling place. Of the data reported, 90 percent of the punch card jurisdictions in this country utilize a central-count tallying process. On the other hand 61 percent of optical scan jurisdictions use a precinct-based tallying process. Poll Workers For reporting jurisdictions, there were at least 845,962 poll workers that worked at polling places on Election Day, which constituted almost one in 200 of the CVAP. There were an average of 5.7 poll workers per precinct and 7.9 per polling place. However, we noted that some jurisdictions use shifts of poll workers while others require poll workers to work the entire Election Day. Jurisdictions reported that 5,252 polling places or precincts were inadequately staffed on Election Day, or 5.8 percent of polling places. Inadequate staffing was reportedly concentrated in four states: Louisiana (64.7 percent of polling places inadequately staffed), Hawaii (44.3 percent), Delaware (28.3 percent), and Illinois (18.4 percent). Most other states reported 7.5 percent or fewer polling places with inadequate staffing. Patterns of inadequate staffing were greatly confounded by the concentration of inadequate staffing numbers in these four states, particularly in the larger states of Illinois and Louisiana. Jurisdictions with higher levels of income and education reported a higher average number of poll workers per polling place or precinct and reported lower rates of staffing problems per precinct. Staffing problems appeared to be particularly acute for jurisdictions in the lowest income and education categories. Small, rural jurisdictions and large urban jurisdictions tended to report higher rates of inadequate numbers of poll workers within polling places or precincts. Predominantly non-hispanic Black jurisdictions reported a greater percentage of polling places or precincts with an inadequate number of poll workers. Predominantly non-hispanic Native American jurisdictions reported the second-highest percentage of staffing problems. This appears to be related to similar higher reports on inadequate numbers of poll workers for Section 5 covered jurisdictions, though at least some of the observed relationships are attributable to Louisiana. Jurisdictions that anticipated Election Day needs reported higher averages of staffing of polling places or precincts and fewer instances of not being able to adequately staff polling places or precincts. For example, jurisdictions in battleground states reported fewer polling places and precincts with inadequate staffing, as did jurisdictions that allow Election Day registration. Jurisdictions with no excuse absentee balloting and those with early voting reported lower rates of

14 2004 Election Day Survey Report Executive Summary, Page 8 September 27, 2005 problems staffing polling places or precincts, perhaps because these alternative modes of voting reduced the Election Day burden for these jurisdictions. Polling Places For reporting jurisdictions, there were at least 174,252 precincts and 113,754 polling places, for an average ratio of 1.45:1 polling places to precincts. There are fewer polling places than precincts due to the administrative practice of consolidating multiple precincts into one polling place. In urban areas precinct consolidation is easier, and perhaps necessary, due to limited availability of suitable locations for polling places in dense population areas. We found higher reported ratios of precincts to polling places in urban areas, and by a consequence, in states and regions with larger urban populations. Other tabulations associated with urban/rural character, such as vote for presidential winner, report similar relationships. Income and education of a jurisdiction are also related, with higher reported ratios of precincts to polling places at higher levels of education and income. For some states, pressures are relieved in Election Day polling places through other methods of voting. Oregon, which conducts its election entirely by mail, has the need for one polling place per county. States with Election Day registration also consolidate fewer precincts than those without, perhaps to aid in the processing of voters at the polls on Election Day. Excluding Oregon, the strongest reported relationship between average voter registration per polling place is found among the population size of the jurisdiction. Jurisdictions of smaller size report a smaller number of registered voters per polling place. This size of the jurisdiction was related to other tabulations, such as the urban/rural character of the jurisdiction, the region the jurisdiction is located in, the type of equipment used, and the presidential winner of the jurisdiction. There is also a relationship between income and education, with lower reported average voter registration per jurisdiction for lower levels of income and education. There is also a relationship between service demands and average registration per polling place, as those jurisdictions with Election Day registration have less registration per polling place than other jurisdictions and those with early voting report higher average registration per polling place. Disability The most significant issue in this chapter is the overall lack of data. Only 26 of the 55 states and territories provided information on disability in response to question 21. While a greater number of polling places were reported to be wheelchair accessible (question 21a), the much smaller numbers of polling places reported to be available to the visually impaired (question 21b) or physically disabled (question 21c) may have resulted from how the survey questions were worded. Some states reported that they interpreted the last two questions as seeking information on the voting equipment in use and its accessibility, rather than the physical configuration of the polling place. Overall, 94.0 percent of the polling places and 70.9 percent of the precincts in this nation were reported to be wheelchair accessible. However, this information reflects data from only half of the nation s election jurisdictions. Fewer than a quarter of the precincts and only 30 percent of the polling places were reported by the states as being locations where a visually impaired voter could cast a ballot in private. Part of the reason this information is so low is that a number of states reported actual zeros in the data cells, rather than leaving them blank. If one eliminates the zeros

15 2004 Election Day Survey Report Executive Summary, Page 9 September 27, 2005 from being part of the calculation, then the percentage rises to 68 percent for the visually impaired data. A physically disabled voter could cast a ballot on an accessible voting system in only about half the precincts and slightly more than 73 percent of the polling places. Common Patterns Across the analysis of separate sections of the Election Day Survey, consistent patterns emerge: Jurisdictions with low education and income, compared with other jurisdictions, tend to report more inactive voter registration, lower voter turnout, higher number of provisional ballots cast, higher drop-off and associated components of overvotes and undervotes, lower average number of poll workers per polling place, and greater percentage of inadequately staffed polling places. While these patterns present a challenge to election administrators, they are consistent with a large body of academic literature that equates higher levels of civic participation to higher levels of education and income. Thus, these findings give us confidence in the overall validity of the responses provided to the EAC on the Election Day Survey and in other patterns we observe. Jurisdictions in states with statewide voter registration databases tend to report less inactive voter registration, higher return rates of absentee ballots, and fewer provisional ballots cast. This suggests that better administration of registration rolls can improve the administration of elections, and perhaps reduce costs by reducing the number of absentee ballots sent to wrong addresses and the number of provisional ballots processed. Jurisdictions covered by the Section 203 of the Voting Rights Act tended to report more inactive voter registration, lower voter turnout, fewer returned absentee ballots, and much greater numbers of provisional ballots cast. These patterns were often similar to those found among predominantly Hispanic and predominantly non-hispanic Native American jurisdictions. These findings appear to be consistent with voters within these jurisdictions having difficulty in navigating the electoral process in a language that is not their native tongue.

16 PART 1 INTRODUCTION Under the Help America Vote Act of 2002 (HAVA), the U.S. Election Assistance Commission (EAC) developed and distributed three surveys to state election directors to obtain baseline election administration data for identifying and prioritizing issues that affect voter enfranchisement and participation in the electoral process. The three surveys are the National Voter Registration Act (NVRA), Election Day, and Military and Overseas Absentee Ballot surveys. This is a report of the Election Day Survey, which is the largest and most comprehensive survey of voting and election administration practices ever conducted by a U.S. governmental organization. The survey was an attempt to create a complete enumeration of voting statistics and election practices in all 50 states, the District of Columbia, and four territories Guam, Puerto Rico, American Samoa, and the U.S. Virgin Islands. State respondents to the survey have reported that 121,862,329 of 177,265,030 registered voters participated in the 2004 general election. This is the highest number of persons to have voted in an election in the United States and an increase of over 14 million voters from the 2000 general election. As a percentage of the citizen voting age population (CVAP) the turnout rate in the 2004 election was 60.7 percent, which increased from 55 percent for the 2000 election and was the highest percentage of turnout since the 1968 election.

17 2004 Election Day Survey Report, Part 1 Introduction, page 2 September 27, 2005 Survey Design and Development Development Timeline In July 2004, the EAC asked Election Data Services Inc. to compile a comprehensive list of data elements for a proposed election administration database. The list of recommended data items included voter registration and voter turnout statistics, election returns for federal offices, information on voting systems and system manufacturers, and organizational information for state and local election jurisdictions. In August 2004, Election Data Services was contracted by the EAC to conduct a telephone survey to determine which data elements state election directors were planning to collect from the November 2004 general election. Results of the telephone survey were presented to the EAC in September. EAC staff then proceeded with the design of the Election Day Survey, which was distributed to state election directors and secretaries of state on October 25, The survey was distributed in an electronic format with a request for a response by January 1, On January 10, 2005, the EAC published a request for proposal for assistance with the analysis and interpretation of the three HAVA surveys, including the Election Day Survey. A contract for survey analysis support was issued on February 15 to Election Data Services Inc., the successful bidder. Work covered by the contract included the tabulation of survey responses, cleanup and clarification of the survey data, analysis and interpretation of survey results, development of recommendations on future data collection, and compilation of the survey results and recommendations in a report to the EAC. The project team providing survey analysis support to the EAC was composed of a principal investigator, a statistical consultant, and support personnel. The principal investigator was Kimball W. Brace, president of Election Data Services, Inc. The statistical consultant was Dr. Michael P. McDonald, an assistant professor of government and politics in the Department of Public and International Affairs at George Mason University. Support personal included research analysts, database programmers, and administrative assistants, all employees of Election Data Services. Survey Questions The Election Day Survey consisted of 24 questions on five major topics: voter registration, election results, voting equipment, poll workers, and voting jurisdictions. The survey questions were as follows:

18 2004 Election Day Survey Report, Part 1 Introduction, page 3 September 27, 2005 Election Day Survey Questions Voter Registration 1. Number of active registered voters (1a), and inactive registered voters (1b). Election Results 2. Number of ballots counted statewide (2a), and by local jurisdiction (2b). 3. Number of ballots cast at polling places on Election Day statewide (3a), and by local jurisdiction (3b). 4. Number of absentee ballots requested statewide (4a), and by local jurisdiction (4b). 5. Number of absentee ballots returned statewide (5a), and by local jurisdiction (5b). 6. Number of absentee ballots counted statewide (6a), and by local jurisdiction (6b); number of absentee ballots not counted (6c); and five most common reasons for rejecting absentee ballots (6d). 7. Whether the state conducts early voting (7a); and number of early ballots counted statewide (7b), and by local jurisdiction (7c). 8. Number of provisional ballots cast statewide (8a), and by local jurisdiction (8b). 9. Number of provisional ballots counted statewide (9a), and by local jurisdiction (9b); and five most common reasons for rejecting provisional ballots (9c). 10. Number of undervotes in each federal contest, by local jurisdiction. 11. Number of overvotes in each federal contest, by local jurisdiction. 12. Number of votes cast for all candidates in each federal contest, by local jurisdiction. Voting Equipment 13. Type and manufacturer of voting systems in use; number of units for each system; software versions, if applicable; and whether used previously in a federal election, by local jurisdiction. 14. Where any of the following voting machine malfunctions occurred, by local jurisdictions and precinct, and whether the affected machines were returned to service: (14a) power failure, (14b) broken counter, (14c) computer failure, (14d) printer failure, (14e) screen failure, (14f) fatal damage to machine, (14g) modem failure, (14h) scanner failure, (14i) ballot encoder or activator failure, (14j) audio ballot failure, and (14k) other malfunctions. Poll Workers 15. Number of poll workers statewide (15a), and by local jurisdiction (15b). 16. Required number of poll workers per precinct or polling place, by law or regulation. 17a. Number of precincts or polling places in each local jurisdiction that did not have the required number of poll workers. 17b. Number of additional poll workers that would have been needed to meet the requirement in question 16 for each precinct that had a deficit of poll workers.

19 2004 Election Day Survey Report, Part 1 Introduction, page 4 September 27, 2005 Election Day Survey Questions (cont.) Voting Jurisdictions 18. What constitutes a local election jurisdiction e.g., county, township, or city? 19. Number of precincts, by local jurisdiction. 20. Number of polling places, by local jurisdiction. 21a. Number of polling places in each local jurisdiction that are wheelchair accessible. 21b. Number of polling places in each local jurisdiction where visually impaired voters can cast a private ballot. 21c. Number of polling places that have an accessible voting system for physically disabled voters. Sources of Information 22. Number of local election jurisdictions providing information for the survey. 23. Contact information for officials of local election jurisdictions. 24. Other sources of information for the survey. Applicability The survey covered 6,568 local election jurisdictions in the 50 states, the District of Columbia, and the four territories. The 24 survey questions were not applicable to all respondents. For example, North Dakota does not have voter registration. Six states with Election Day registration are exempt from provisional balloting. Questions on voting equipment would not be applicable to jurisdictions that use hand-counted paper ballots. Survey Response Election Jurisdictions Although the Election Day Survey was distributed to 55 state election directors (including four territories and the District of Columbia), the state directors were charged with gathering information from large numbers of local election jurisdictions to complete the survey. Texas has 254 counties, and Wisconsin has some 1,910 municipalities that conduct elections. The 6,568 election jurisdictions represented in the EAC survey database include 3,090 counties and county equivalents, and 3,460 cities and towns in Wisconsin and the six New England states. Some 1,500 municipalities in Michigan and 2,600 municipalities in Minnesota also conduct elections; however, only county-level information was obtained from Michigan and Minnesota for the survey.

20 2004 Election Day Survey Report, Part 1 Introduction, page 5 September 27, 2005 Number of Local Election Jurisdictions Alabama 67 Nevada 17 Alaska 1 New Hampshire 242 Arizona 15 New Jersey 21 Arkansas 75 New Mexico 33 California 58 New York 58 Colorado 64 North Carolina 100 Connecticut 169 North Dakota 53 Delaware 3 Ohio 88 District of Columbia 1 Oklahoma 77 Florida 67 Oregon 36 Georgia 159 Pennsylvania 67 Hawaii 5 Rhode Island 39 Idaho 44 South Carolina 46 Illinois 110 South Dakota 66 Indiana 92 Tennessee 95 Iowa 99 Texas 254 Kansas 105 Utah 29 Kentucky 120 Vermont 246 Louisiana 64 Virginia 134 Maine 517 Washington 39 Maryland 24 West Virginia 55 Massachusetts 351 Wisconsin 1,910 Michigan 83 Wyoming 23 Minnesota 87 American Samoa 1 Mississippi 82 Guam 1 Missouri 116 Puerto Rico 110 Montana 56 Virgin Islands 1 Nebraska 93 Total 6,568 Coverage At the time the contract for survey analysis support was issued in mid-february, the EAC had received responses to the Election Day Survey from 48 states and territories. By March 17, most state responses had been received. The last state responses were added to the database on March 31 (Rhode Island) and April 13 (Michigan). As of April 15, there were two nonrespondents to the Election Day Survey: Guam and American Samoa. Supplemental data was added to the database up to July 15, This included corrected data from a follow-up review of survey data that was conducted during the second week of July. On July 8, 2005, a spreadsheet containing data tabulated for local election jurisdictions was sent to each state election director. The state directors were asked to review the spreadsheets, provide missing data, correct data entries, if necessary, and return the spreadsheets by July 15, Responses to the

21 2004 Election Day Survey Report, Part 1 Introduction, page 6 September 27, 2005 survey follow-up review were received from 26 states. As of September 1, Guam and American Samoa were still nonrespondents to the Election Day Survey. Even with the follow-up review, many responses to the Election Day Survey are incomplete. In some cases, responses are missing one or more local election jurisdictions. In other cases, a response is missing for certain questions for example, question 14 on voting equipment malfunctions. On March 15, the original cutoff date for data tabulation, overall completeness rates for original state responses varied from 91.5 percent complete to less than 20 percent, as shown below: No. States Survey Completeness Rates by Mar. 15 Over 80 percent complete 5 60 to 80 percent complete to 60 percent complete to 40 percent complete 6 Less than 20 percent complete 2 No response* 3 *As of March 15, only statewide data had been received for the state of Rhode Island. Data for local election jurisdictions in Rhode Island was added to the survey database on March 31. Many states provided supplemental data in response to requests for missing data or clarifications of problem data. Some data had not been reported consistently. For example, two-thirds of the nation s jurisdictions provided responses to questions on active registrations, but for inactive registration, less than half reported data. By comparing survey responses with reported registration data, Election Data Services determined that 20 states combined active and inactive registrations in their counts of overall registrations in the state. Twenty-six states reported only active registration. In four states, the determination of whether to report active and inactive voters in voter registration totals is at the discretion of individual local jurisdictions. Responses to other election data on the number of ballots cast by mode of voting, absentee ballots, provisional voting, and the number of undervotes by federal office were often incomplete. The following table summarizes the coverage of state responses to selected questions on the survey as of July 15, This summary includes supplemental data provided by state election directors as a result of the state follow-up review. Number of Reponses Coverage Rate Individual Survey Questions (Jurisdictions) (Percent) 1a. Active registration 4, b. Inactive registration 3, a. Ballots counted 6, a. Ballots cast on Election Day 3, a. Absentee ballots requested 4, a. Absentee ballots returned 4, a. Absentee ballots counted 4, c. Absentee ballots not counted 1, b. Early ballots counted 1,

22 2004 Election Day Survey Report, Part 1 Introduction, page 7 September 27, 2005 Coverage Rate Individual Survey Questions (Percent) 8a. Provisional ballots cast 3, a. Provisional ballots counted 2, a. Votes cast for President 6, a. Presidential undervotes 4, a. Presidential overvotes 1, b. Votes cast for U.S. Senator 4, b. Senate undervotes 3, b. Senate overvotes c. Votes cast for U.S. Representative 6, c. Congressional undervotes 4, c. Congressional overvotes Type of voting equipment n/a* n/a* 14. Voting equipment malfunctions n/a* n/a* 15a. Number of poll workers 4, Required number of poll workers per precinct 1, a. Precincts with fewer poll workers than required 2, Number of precincts 5, Number of polling places 5, a. Wheelchair-accessible polling places 3, b. Polling places where visually impaired cast private ballots c. Polling places with accessible voting systems 1, *Coverage rates could not be calculated for questions on voting equipment because many jurisdictions provided data for more than one type of voting equipment. From conversations with state election directors and an examination of survey responses, Election Data Services determined that some state election directors sent the survey or selected questions from the survey directly to local jurisdictions, while others rewrote the questions on a new version of the survey that was distributed to local election jurisdictions. In several instances, election directors noted in their responses that local election officials had not carefully read or fully understood certain questions on the Election Day Survey. For example, some responses to question 3b, ballots cast on Election Day, were identical to question 2b, total number of ballots counted, and did not exclude absentee ballot totals. Some local officials may have interpreted ballots cast (question 3b) as individual pages of a multipage ballot; therefore, five voters casting a three-page ballot would have been interpreted as 15 ballots cast, rather than five ballots cast. There are data quality issues, cases of missing data, and, inevitably, data entry errors. Some data entry and reporting errors were detected by data integrity reports that identified rates in excess of 100 percent for example, more ballots counted than registered voters, more ballots counted than ballots cast, more absentee ballots returned than absentee ballots requested, or more provisional ballots counted than provisional ballots cast. Other errors were detected by reports that compiled the highest and lowest 15 counties for each data category or reports that compared the survey responses with other data sources, such as certified election data published on the Web or surveys on similar topics conducted by other election organizations, such as electionline.org and the National

23 2004 Election Day Survey Report, Part 1 Introduction, page 8 September 27, 2005 Association of Secretaries of State. Still other errors were corrected through telephone calls for data clarifications e.g., extra digits (123 ballots cast, instead of 1,023 ballots) or transpositions (113 provisional ballots counted, instead of 131 ballots). Some states were asked to provide corrected data when, for example, (1) the number of absentee ballots returned was higher than the number of ballots requested (more than 100 jurisdictions), (2) the number of absentee ballots counted was higher than the number of ballots returned (more than 140 jurisdictions), or (3) the number of provisional ballots counted was higher than the number of ballots cast (15 jurisdictions). Election Data Services has attempted to locate and correct errors in larger magnitude, but we are certain that smaller errors exist in the data. It is hoped that these small errors will not undermine the results of the analysis that we report, and that these errors are minimized when data is aggregated to the county level. Some problems remain with the data because requests for data corrections have not yet been received. Note: Because of the data quality issues, it is important to check the primary data sources (i.e., original survey responses) if certain items in this report seem questionable. Data errors were discovered and corrected throughout the analysis and report-writing phases of this project, right up to the date of the final report. Data Tabulation The electronic format that the EAC chose for the Election Day Survey was a Microsoft Excel spreadsheet. While most states used the Excel format for their responses, there were considerable variations among the states in the presentation of data presented on the Excel spreadsheets. For example, while most states reported election jurisdiction data in rows and survey questions in columns, a few states used the opposite format i.e., rows for questions, and columns for jurisdictions. In some instances, particularly on voting equipment, individual data cells contained responses to two or more questions e.g., equipment manufacturer and equipment name or type and software version. Some survey responses or supplements to survey responses were provided in Microsoft Word documents or.csv,.pdf, or html files (i.e., comma delimited text, Portable Document Format, or Hypertext Markup Language Web documents). To tabulate the survey, Election Data Services standardized the survey responses in new Microsoft Excel import files for addition to a special SQL (Structured Query Language) relational database that was created for the project. The special EAC database for the Election Day Survey has 6,568 records (one record for each local election jurisdiction) and 70 columns of data. Data integrity and quality assurance reports to assess jurisdictional coverage and data quality issues were produced from this database as well as 14 tables that present the survey results and form the basis for this report. Each of the 14 tables has a separate chapter in this report. The tables are entitled as follows: Table 1. Population Estimates Table 2. Voter Registration Table 3. Ballots Counted Table 4. Turnout Source Table 5. Absentee Ballots Table 6. Provisional Ballots Table 7. Drop-Off Table 8. Overvotes and Undervotes Table 9. Voting Equipment Usage Table 10. Voting Machines Table 11. Voting Equipment Malfunctions Table 12. Poll Workers Table 13. Polling Places Table 14. Disability

24 2004 Election Day Survey Report, Part 1 Introduction, page 9 September 27, 2005 Data Measurement Four basic methodologies were used to analyze the results of the Election Day Survey. They are (1) data entry and tabulation checks to provide complete and consistent nationwide coverage; (2) calculation of rates and ratios to provide meaningful comparisons among states and counties; (3) cross-tabulation and correlation by different criteria e.g., type of voting equipment used to reveal patterns between two variables; and (4) regression analysis to provide a statistically rigorous analysis of patterns revealed through cross-tabulation and correlation. Table Format The 14 tables present statewide summaries of the survey results. The questions are in columns and state responses to the survey questions are in rows. Next to each column containing a response to a survey question or a calculation representing responses to two or more questions is a column labeled Cases. The Cases column provides information on the number of jurisdictions that are represented by the survey response or calculation. For example, in the following illustration from Table 2, state responses to question 1a of the Election Day Survey on the number of active registered voters are in column 6, Active Registration. In column 7, Cases shows the number of local election jurisdictions covered by the state s response to survey question 1a. Column 8, Percent Active Registration lists the number of active registered voters in column 6, divided by Reported Total Registration in column 4. Cases in column 9 shows the number of local jurisdictions that responded to question 1a on active registered voters (col. 6) and the number of jurisdictions covered by the calculation of total registered voters in column 4.

25 2004 Election Day Survey Report, Part 1 Introduction, page 10 September 27, 2005 Moving down the table and following the 55 state responses is a series of calculations showing the nationwide totals, with maximum, minimum, and average counts for each variable. These calculations are followed by a series of cross-tabulations for interpreting the survey data. The crosstabulations are the same for each table and cover the following three categories: (1) election administration factors, (2) geographic and demographic factors, and (3) political factors: Election Administration Geo/Demographic Political Type of Voting Equipment Regions Battleground States Changed Voting Equipment since 2000 Statewide Voter Registration Database Election Day Registration Provisional Ballot Acceptance No-Excuse Absentee Balloting Early Voting Sec. 203 Language Minority Requirements Sec. 5 Preclearance of Voting Procedures Urban to Rural Size of Jurisdiction Race and Ethnicity Median Income High School Education Presidential Margin of Victory Red versus Blue Jurisdictions Cross Tabulations The following is a description of the cross-tabulation factors for interpreting the survey results. Type of Voting Equipment Cross-tabulations by voting equipment are for five generic equipment types: (1) paper, (2) lever, (3) electronic, (4) punch card, and (5) optical scan. Paper refers to the Australian or mark choice ballot, on which voters choose candidates or responses to ballot questions by marking boxes on a paper ballot, which are then counted by hand. Lever refers to mechanical lever machines, which display a full-face ballot with a small lever next to each candidate s name and each ballot question. Voters enter a curtain-enclosed booth, make their choices for candidates and ballot questions by flipping the small levers, and then pull a large lever to open the curtain to exit the booth and record their votes on counters located on the back the machines. No paper trail of an individual voter s choices are ever produced on a lever machine. Electronic refers to Direct Record Electronic (DRE) systems where voters use push buttons, select wheels, or touch screens to choose candidates or responses to ballot questions. Their choices are recorded and tabulated electronically in removable memory components. Punch card refers to both Votomatic and DataVote style systems, where voters insert paper ballot cards into a device and punch out chads next to candidate names and ballot questions. 1 The voted punch cards are then 1 A Votomatic ballot card is prescored and printed only with numbered voting positions. A Votomatic ballot card is inserted into a frame on which a booklet identifying candidates or answers to ballot questions has been attached. A stylus is used to punch out chads at various voting positions. A DataVote ballot card is printed with a candidate name or answer to a ballot question at each voting position. A DataVote ballot card is inserted into a frame fitted with a movable device similar to a one-hole punch for punching out chads at voting positions. Because the candidate names are printed on the card, most ballots provided to voters encompass multiple physical cards.

26 2004 Election Day Survey Report, Part 1 Introduction, page 11 September 27, 2005 processed by a computer reader, which tabulates the vote. Optical scan systems use paper ballots on which voters mark their choices for candidates and ballot questions with pencils or pencil-like devices. Voted ballot cards are then scanned by machines using marksense (e.g., infrared) technology to tabulate the vote. The data source for the voting equipment cross-tabulations is survey question 13, which asked for the type and manufacturer of voting equipment used at the November 2004 election. The number of local jurisdictions for each equipment type is shown below. The unknown category is for jurisdictions that did not respond to the survey question. Type of Voting Equipment Jurisdictions Paper 1,734 Lever 394 Electronic 608 Punch Card 260 Optical Scan 2,541 Multiple Systems 123 Unknown 908 Changed Voting Equipment Since 2000 The cross tabulation for voting equipment changes are based on responses to survey question 13 and information maintained by Election Data Services on voting equipment used at the November 2000 general election. The number of local jurisdictions that used different voting equipment in November 2000 is as follows: Changed Voting Equipment Since Nov Jurisdictions Yes 1,753 No 4,815 Statewide Voter Registration Database Section 303 of HAVA requires states to implement a statewide voter registration database by January 1, 2004, unless a waiver was obtained to extend the implementation deadline to January 1, Seventeen states had statewide databases in place for the November 2004 general election, and the number of local jurisdictions in those states is as follows: Jurisdictions in states with statewide voter registration databases in place for the November 2004 election: 1,335 Alaska Hawaii New Mexico Arizona Kentucky Oklahoma Connecticut Louisiana South Carolina Delaware Massachusetts South Dakota Dist. of Columbia Michigan West Virginia Georgia Minnesota Jurisdictions in other states: 5,233 Election Day Registration Six states allow persons to register and vote on Election Day. 2 Proponents of Election Day Registration (EDR), also called same-day voter registration, maintain that EDR increases the 2 Another state, Rhode Island, is not considered an Election Day Registration (EDR) state, but allows persons to register on Election Day to vote for president only.

27 2004 Election Day Survey Report, Part 1 Introduction, page 12 September 27, 2005 opportunity to cast a vote and that the EDR states have higher than average voter turnout rates. The number of local jurisdictions in states that have EDR is as follows: Jurisdictions in states (6) with Election Day registration: 2,823 Idaho Minnesota Wisconsin Maine New Hampshire Wyoming Jurisdictions in other states: 3,745 Provisional Ballot Acceptance Provisional balloting allows voters who believe that they registered to cast a ballot even though their names do not appear on a voter list. A provisional ballot may be counted if, after investigation, it is determined that the voter was, in fact, eligible to vote. Provisional balloting is mandated by HAVA, although many states already had provisional balloting or other fail-safe voting procedures before HAVA was enacted. However, provisional balloting procedures differ among the states, and one major difference is where provisional ballots are cast. Provisional ballots in 28 states are disqualified if cast outside the voter s home precinct, while in 18 states, provisional ballots are eligible to be counted if cast in the voter s home jurisdiction e.g., county or municipality but not necessarily in the voter s home precinct. The number of local jurisdictions in states with in-precinct and out-of-precinct rules for counting provisional ballots is shown below. Other local jurisdictions are in states that have no such rules or are exempt from HAVA s provisional balloting requirement. HAVA exempts states that do not have voter registration and states that have Election Day registration, although three states with Election Day registration, Maine, Wisconsin and Wyoming, use provisional ballots for first-time voters whose names do not appear on voter lists and who do not have proper identification at the polls on Election Day. Jurisdictions in states (28) where provisional ballots must be cast in the voter s home precinct (in precinct only): 4,350 Alabama Kentucky Ohio Connecticut Massachusetts Oklahoma Dist. Columbia Michigan South Carolina Florida Mississippi South Dakota Hawaii Missouri Tennessee Indiana Montana Texas Iowa Nebraska Virginia Kansas Nevada West Virginia New Jersey Wisconsin New York Wyoming Jurisdictions in states (18) where provisional ballots are eligible to be counted if cast in the voter s home jurisdiction but not necessarily in the voter s home precinct 1,162 (anywhere in jurisdiction): Alaska Georgia Oregon Arizona Illinois Pennsylvania Arkansas Louisiana Rhode Island California Maryland Utah Colorado New Mexico Vermont Delaware North Carolina Washington Jurisdictions in other states: 1,056

28 2004 Election Day Survey Report, Part 1 Introduction, page 13 September 27, 2005 No Excuse Absentee Balloting The EAC defined absentee voting as voting prior to Election Day which requires that the voter meet qualifications other than those generally required to register to vote. For example, a voter might have to attest that he or she will be absent from the voting jurisdiction on Election Day. Many states now allow voters to cast absentee ballots without conditions. Cross-tabulations by no-excuse absentee balloting apply to jurisdictions in the following 24 states: Jurisdictions in states (24) with no-excuse absentee ballots: 3,781 Alaska Kansas North Dakota Arizona Louisiana Oklahoma California Maine South Dakota Colorado Montana Utah Florida Nebraska Vermont Hawaii Nevada Washington Idaho New Mexico Wisconsin Iowa North Carolina Wyoming Jurisdictions in other states: 2,787 Early Voting The EAC defined early voting as any voting that occurred prior to November 2, 2004, for which there were no eligibility requirements. For example, the voter did not have to attest that he/she would be absent from the voting jurisdiction on the day of the election. The number of local jurisdictions in the 27 states that conduct early voting is as follows: Jurisdictions in states (27) with early voting: 1,701 Alaska Indiana North Dakota Arizona Iowa Oklahoma Arkansas Kansas South Dakota California Maine Tennessee Colorado Montana Texas Florida Nebraska Utah Georgia Nevada Vermont Hawaii New Mexico West Virginia Idaho North Carolina Wyoming Jurisdictions in other states: 4,867 Section 203 Language Minority Requirements Section 203 of the Voting Rights Act requires election jurisdictions to provide language assistance at the polls, such as translation services or special ballots, if a language minority group represents a certain proportion of voting age citizens. Covered language minority groups are American Indians, Asian Americans, Alaskan Natives, and Spanish-heritage citizens. Section 203 cross-tabulations apply to 468 jurisdictions in 27 states. 3 3 Implementation of the Provisions of the Voting Rights Act Regarding Language Minority Groups, 28 C.F.R. Part 55.

29 2004 Election Day Survey Report, Part 1 Introduction, page 14 September 27, 2005 Section 5 Preclearance of Voting Procedures Section 5 of the Voting Rights Act requires certain jurisdictions to obtain federal approval ( preclearance ) before implementing changes to voting procedures. Section 5 cross-tabulations apply to 880 covered jurisdictions in 16 states. 4 Regions Cross-tabulations by geographic area apply to four regional groupings of states used by the U.S. Census Bureau. These groupings exclude the four territories. The number of local jurisdictions in each of the four census regions Northeast, Midwest, South, and West is as follows: Jurisdictions in the nine Northeastern states: 1,710 Connecticut New Hampshire Pennsylvania Maine New Jersey Rhode Island Massachusetts New York Vermont Jurisdictions in the 12 Midwestern states: 2,902 Illinois Michigan North Dakota Indiana Minnesota Ohio Iowa Missouri South Dakota Kansas Nebraska Wisconsin Jurisdictions in the 17 Southern states: 1,423 Alabama Kentucky South Carolina Arkansas Louisiana Tennessee Delaware Maryland Texas Dist. of Columbia Mississippi Virginia Florida North Carolina West Virginia Georgia Oklahoma Jurisdictions in the 13 Western states: 420 Alaska Idaho Utah Arizona Montana Washington California Nevada Wyoming Colorado New Mexico Hawaii Oregon Jurisdictions in four territories: 113 Urban to Rural Cross-tabulations by population density for four area types urban, suburban, small town, and rural were created for this study from the U.S. Census P.L Redistricting Data Summary File. The area quartiles were created by dividing the populations of geographic units represented in the P.L. File by the areas of those units in square miles. The four territories are not covered by these calculations. The number of local jurisdictions in each population density quartile is as follows: Area Population Density Jurisdictions Urban 1,000 people per square mile or more 567 Suburban 250 to 999 people per square mile 871 Small Town 50 to 249 people per square mile 1,710 Rural 0 to 49 people per square mile 3,307 Territories (not available) Procedures for the Administration of Section 5 of the Voting Rights Act of 1965, As Amended, 28 C.F.R. Part 51, and Appendix to Part 51, Jurisdictions Covered Under Section 4(b) of the Voting Rights Act, as Amended.

30 2004 Election Day Survey Report, Part 1 Introduction, page 15 September 27, 2005 Size of Jurisdiction (VAP) Cross-tabulations by size of jurisdiction are based on selected ranges of the estimated voting age population (VAP) for the November 2, 2004, general election. VAP is defined as all persons age 18 and older residing within a jurisdiction. Estimated VAP for November 2004 is based on U.S. Census Bureau estimates of the population by age on July 1, 2002, and July 1, The November 2004 estimated VAP was constructed by extrapolating forward the difference between the July 1, 2002, and July 1, 2003, census estimates. The four territories are not covered by these ranges. The number of local jurisdictions in each range is as follows. Voting Age Population (VAP) Jurisdictions Less than 1,000 1,761 1,000 to 3,499 1,165 3,500 to 9,999 1,043 10,000 to 49,999 1,704 50,000 to 249, ,000 to 999, ,000,000 or more 25 Territories and problem jurisdictions (not available) 144 Most of the small jurisdictions are municipalities in Wisconsin and the six New England states. Cross-tabulations based on voter registration would have been preferable for this study but could not be created because of a lack of consistent voter registration data. Race and Ethnicity Cross-tabulations by race and ethnicity are also based on population counts from the U.S. Census P.L Redistricting Data Summary File for persons 18 years and over as well as Hispanic/Latino and non-hispanic/latino persons by race (63 categories). Race and Ethnic Categories Jurisdictions Predominantly Non-Hispanic White 6,284 Predominantly Non-Hispanic Black 85 Predominantly Non-Hispanic Native American 24 Predominantly Hispanic 50 Territories and problem jurisdictions (not available) 145 Median Income Cross-tabulations by median income are based on income data in the U.S. Census Summary File 1 (SF 1). The four territories are not included in these tabulations. The number of local jurisdictions in each range by median income is as follows: Income Categories Jurisdictions Less than $25, $25,000 to 29, $30,000 to 34,999 1,372 $35,000 to 39,999 1,215 $40,000 to 44, $45,000 to 49, $50,000 or more 1,180 Territories and problem jurisdictions (not available) 151

31 2004 Election Day Survey Report, Part 1 Introduction, page 16 September 27, 2005 High School Education Cross-tabulations by high school graduation or equivalent diploma are based on educational attainment data in the U.S. Census Summary File 1 (SF 1). The four territories are not included in these tabulations. The number of local jurisdictions in each range by educational attainment is as follows: High School Graduation Rates Jurisdictions Less than 60 percent to 70 percent to 80 percent 1, to 90 percent 3, percent or higher 873 Territories and problem jurisdictions (not available) 151 Battleground States Cross-tabulations by battleground state apply to the 2004 presidential election and are based on the number of local jurisdictions in the 17 battleground states as follows: Jurisdictions in the 17 battleground states: 3,093 Arkansas Minnesota Oregon Arizona Missouri Pennsylvania Colorado Nevada Washington Florida New Hampshire West Virginia Iowa New Mexico Wisconsin Michigan Ohio Jurisdictions in other states: 3,475 Presidential Margin of Victory Cross-tabulations by margin of victory are for the 2004 presidential election by the following quintiles: Less than 2.5 percent, 2.5 to 5.0 percent, 5.0 to 7.5 percent, 7.5 percent to 10.0 percent, and 10 percent or more. The number of local jurisdictions in each margin of victory quintile is shown below. The number does not sum to 6,568, the number of jurisdictions covered by the survey, because election returns were not reported for some smaller jurisdictions whose votes are included in the totals of another jurisdiction. Presidential Margin of Victory Jurisdictions Less than 2.5 percent percent to 5.0 percent percent to 7.5 percent percent to 10.0 percent percent or more 4,492

32 2004 Election Day Survey Report, Part 1 Introduction, page 17 September 27, 2005 Red versus Blue Jurisdictions Cross-tabulations by red versus blue apply to the 2004 presidential election results and are based on local jurisdictions won by John Kerry (blue) and George W. Bush (red). The number of jurisdictions by margin of victory is shown below. The number does not sum to 6,568, the number of jurisdictions covered by the survey, because election returns were not reported for some smaller jurisdictions whose votes are included in the totals of another jurisdiction. Candidate Margin of Victory Jurisdictions Red (Bush) Greater than 55 percent 3,115 Red (Bush) 50 percent to 55 percent 982 Red (Bush) Less than 50 percent 136 Tied 25 Blue (Kerry) Less than 50 percent 150 Blue (Kerry) 50 percent to 55 percent 872 Blue (Kerry) Greater than 55 percent 1,161 REFERENCES [[references aren t cited in text]] Brady, Henry E., Justin Buchler, Matt Jarvis, and John McNulty. Counting All the Votes: The Performance of Voting Technology in the United States. Department of Political Science, Survey Research Center, and Institute for Governmental Studies, University of California, Berkeley, September Dēmos Democracy Program. Election Day Registration Helps America Vote, Winter electionline.org. Solution or Problem? Provisional Ballots in 2004, electionline.org Briefing, March electionline.org. The 2004 Election, electionline.org Briefing, December electionline.org. Statewide Voter Registration Database Status, 5 April 2005 < National Association of Secretaries of State. Early Voting Survey, September National Association of Secretaries of State. Fact Sheet: The Different Types of Voting Systems, June United States. Election Assistance Commission. Glossary for Voting Systems (DRAFT), 12 April 2005 <

33 PART 2 ELECTION DAY SURVEY RESULTS Any discussion about the voting process must take into account several requirements that have been built into the American electoral system. First, not everyone can vote. Persons must have reached a certain age. The 26th Amendment, adopted in 1972, placed the minimum voting age at 18 years. Persons age 18 and older are commonly referred to as the voting age population (VAP). Second, being of voting age is not necessarily a guarantee of voting eligibility, because most states also require persons to be United States citizens. This smaller group of individuals is referred to as the citizen voting age population (CVAP). Third, in most states persons must also register to vote. Some states have made the registration process easier than others, such as the six states that have adopted Election Day Registration. The rural state of North Dakota has no registration requirement. But registering is just another step in the voting process. People have to turn out and cast a ballot. The easing of absentee balloting restrictions and the adoption of mail-in ballots has increased voter turnout to a certain degree. But the fourth step of the process still comes down to people making the effort to vote. Fifth, the American electoral system is unique in that it presents voters with many decisions on Election Day. Not only are there contests among candidates for federal offices, but also for state and local offices. Many states and localities place referendums on the ballot as well. Each contest on the ballot presents voters with a choice of whether to participate. At each step in the voting process, the demographic and political makeup of the electoral body changes. Individuals and various groups of individuals are affected differently. This study of the Election Day Survey analyzes data generated by election systems at each of the five steps. The data is presented in the following series of tables: Table 1. Population Estimates Table 2. Voter Registration Table 3. Ballots Counted Table 4. Turnout Source Table 5. Absentee Ballots Table 6. Provisional Ballots Table 7. Drop-Off Table 8. Overvotes and Undervotes Table 9. Voting Equipment Usage Table 10. Voting Machines Table 11. Voting Equipment Malfunctions Table 12. Poll Workers Table 13. Polling Places Table 14. Disability Each table has a separate chapter that contains background information about survey questions, the historical context for interpreting the survey results, and stipulations about jurisdictional coverage and the applicability of data items. The survey results in each table are summed to the state level for 6,567 local election jurisdictions in the U.S. Election Assistance Commission dataset. The survey results are also subtotaled for each of the 18 cross-tabulation factors described earlier in Part 1. Each chapter provides an analysis of the survey results for each cross-tabulation factor. There is a separate chapter containing recommendations on future data collection.

34 Chapter 1 Population Estimates Table 1 presents estimates of the voting age population (VAP) and the citizen voting age population (CVAP) in the United States for the November 2, 2004, general election. The VAP is defined as all persons age 18 and older residing within a jurisdiction a county, parish, or township, depending on where elections are administered. Estimated VAP for November 2004 is constructed from U.S. Census Bureau population estimates by age and jurisdiction for July 1 of a given year. (Census estimates are available at As of this writing, the July 1, 2004,VAP estimates have not been released, but they are anticipated by the end of the summer of We constructed the November 2, 2004, estimated VAP used in this report by extrapolating forward the difference between the July 1, 2002, and July 1, 2003, census estimates. This method constructs the best approximation of the November 2, 2004, VAP for local election jurisdictions. But we recognize that the method may incorrectly estimate population for a jurisdiction, such as underestimating population growth, particularly for jurisdictions with small populations, such as townships. To construct the best VAP available estimate for townships, we assigned the ratio of the newly released Census Bureau July 1, 2004, total population estimate from the county to the township to apportion our November 2, 2004, county-level VAP estimate to the township. It is important to understand that VAP is not a perfect estimate of those eligible to vote. VAP does not include estimates of voting-eligible persons living overseas. It includes persons who are ineligible to vote under state laws, such as noncitizens; ineligible felons, depending on state law; those determined by a court to be incompetent; those who are not registered to vote; and persons who might have moved recently. But obtaining uniform data for jurisdictions nationwide for each of these circumstances is impossible, and therefore, no possible adjustment can be made to the base data. We can, however, account for noncitizens, the largest ineligible population, by estimating the CVAP. We constructed CVAP by applying the 2000 census estimate of CVAP (which was obtained by Election Data Services as a special tabulation from the Census Bureau) as a percentage of the 2000 census VAP to the November 2, 2004, population estimates described above. This method implicitly assumes that the April 1, 2000, report of the percentage citizens of VAP is equal to the November 4, 2004, percentage citizens of VAP. Methods exist to estimate the other eligible and ineligible populations, such as accounting for the overseas eligible population and ineligible felons from Department of Justice reports (McDonald and Popkin 2001; McDonald 2002), but no sound methodology exists to apportion these populations to counties and townships. For consistency across reporting units, we do not further adjust CVAP to attempt to better measure the voting-eligible population.

35 2004 Election Day Survey Report, Part 2 Survey Results Population Estimates, Page 1-2 September 27, 2005 Applicability and Coverage VAP and CVAP estimates were available for all 50 states, the District of Columbia, and Puerto Rico. The Census Bureau did not produce post-2000 population estimates for Guam, American Samoa, and the Virgin Islands. Population Estimates Table 1 presents population estimates for the analysis of the survey results. Table 1 provides estimates of the VAP and the CVAP, and calculates CVAP as a percentage of VAP. The column headings in Table 1 are as follows: Column Headings for Table 1. Population Estimates Col.- Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question Estimated VAP Estimated November 2004 voting age population (VAP) 5 Cases Number of jurisdictions for which VAP estimates were constructed Estimated. Citizen VAP Estimated November 2004 citizen voting age population (CVAP) 7 Cases Number of jurisdictions for which CVAP estimates were constructed 8 Percent 2004 Citizen of Total VAP Estimated November 2004 CVAP (col. 6) divided by estimated November 2004 total VAP (col. 4) 9 Cases Number of jurisdictions for which VAP and CVAP estimates were constructed 10 Cases > 100% Number of jurisdictions where estimated November 2004 CVAP is greater than total November 2004 VAP

36 2004 Election Day Survey Report, Part 2 Survey Results Population Estimates, Page 1-3 September 27, 2005 Analysis of Estimates The following is our analysis of the data in Table 1 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and statelevel summary of the population data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions Summary The uneven distribution of noncitizens across jurisdictions underscores the importance of using CVAP in addition to VAP when drawing conclusions of survey results across jurisdictions. If VAP were used, rates would be underestimated for jurisdictions with high proportions of noncitizens relative to other jurisdictions. Jurisdictions with high proportions of noncitizens can be found in Western states, particularly California; in urban and small cities; and in Section 203 and Section 5 Voting Rights jurisdictions, among other categories. States Nationally, the 2000 census reported that 92.9 percent of the U.S. voting age population are citizens. The distribution of VAP and CVAP across the states and within cross-tabulations is reported in Table 1. California has the largest voting-age population at 26.6 million. California also has the largest noncitizen population, with only 81.3 percent of the VAP classified as citizens in the 2000 census. West Virginia has the smallest noncitizen population as a percentage of the VAP, with 99.4 percent of the VAP classified as citizens. Regions The largest proportion of non-citizens are located in the West, where only 86.8 percent of VAP are citizens. Jurisdictions located in the Midwest have the highest proportion of citizens, with 96.8 percent. Jurisdictions in the Northeast and the South fall in the middle with 94.3 percent and 93.8 percent citizens, respectively. In all, 224 jurisdictions were reported as having zero noncitizens among the VAP, primarily located in regions outside the West. Urban to Rural Among urban to rural categories, urban jurisdictions have the lowest percentage of citizens, 88.0 percent. The remaining categories fall between 94.4 percent citizen in suburban to 97.3 percent citizen in rural jurisdictions. Size of Jurisdiction The smallest jurisdictions have citizenship of 99.3 percent of VAP. For small- to medium-sized jurisdictions up to 250,000 VAP, citizenship is above 96.1 percent. For the 23 largest jurisdictions in the nation, those with 1,000,000 or more, citizens are 83.2 percent of the VAP.

37 2004 Election Day Survey Report, Part 2 Survey Results Population Estimates, Page 1-4 September 27, 2005 Race and Ethnicity Among racial and ethnic categories, predominantly Hispanic jurisdictions have the lowest percentage of citizens, only 75.7 percent. In predominantly non-hispanic White jurisdictions, greater than 94.0 percent are citizens, and up to 98.0 percent are citizens in predominantly non- Hispanic Native American jurisdictions. Median Income Lower income jurisdictions tend to have higher citizenship rates. The percentage of citizens among the voting age population ranges from 89.7 percent to 96.6 percent. High School Education Jurisdictions with lower percentages of the population completing high school have higher percentages of noncitizens. Jurisdictions with below a 60 percent high school completion rate have an 86.2 percent citizenship rate. Those above 90 percent high school completion report 94.0 percent citizenship among the VAP. Section 203 Language Minority Requirements Jurisdictions covered under Section 203 of the Voting Rights Act have higher percentages of noncitizens. Section 203 jurisdictions are 85.4 percent citizen, while noncovered jurisdictions are 96.6 percent citizen. Section 5 Preclearance of Voting Procedures Jurisdictions covered under Section 5 of the Voting Rights Act have higher percentages of noncitizens. Section 5 jurisdictions are 90.3 percent citizen, while other noncovered jurisdictions are 93.8 percent citizen. Type of Voting Equipment Jurisdictions that use hand-counted paper ballots have the highest proportion of citizens, 98.1 percent, which may be because paper ballots are primarily used by smaller jurisdictions. Crosstabulations with other types of voting equipment are in the low- to mid-90 percent range. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment from the 2000 election have a lower percentage of citizens, 88.7 percent, than other jurisdictions, 94.9 percent. The difference is partially a consequence of voting equipment changes in populous southern California and southern Florida counties with high noncitizen populations. Statewide Voter Registration Database States with statewide voter registration databases have a slightly higher percentage of citizens, 95.3 percent, than those that do not, 92.2 percent. Election Day Registration States with Election Day Registration (EDR) have higher percentages of citizens, 97.2 percent, than those that do not, 92.6 percent. This difference is primarily attributed to the number of EDR states in

38 2004 Election Day Survey Report, Part 2 Survey Results Population Estimates, Page 1-5 September 27, 2005 the Northeast and Midwest. The two Western states with EDR, Idaho and Wyoming, also have high levels of citizenship. Provisional Ballot Acceptance States that allow provisional ballots to be counted if cast outside a voter s home precinct have a lower percentage of citizens, 90.6 percent, than those that accept ballots cast in home precincts only, 94.3 percent, or do not have provisional ballots, 97.0 percent. No Excuse Absentee Balloting States with no excuse absentee balloting have lower percentages of citizens than other states, 90.0 percent versus Early Voting States with early voting have a lower percentage of citizens, 89.7 percent, than states that do not have early voting, 95.7 percent. Battleground States Battleground states in the November 2004 general election had a higher percentage of citizens than those that were not battleground states, 95.0 percent versus 91.7 percent. Presidential Margin of Victory There is no pattern of citizenship among jurisdictions within states by presidential margin of victory. The percentage of citizens ranges from 90.7 percent to 96.0 percent among the categories. Red versus Blue Jurisdictions Jurisdictions won by Bush tend to have higher percentages of citizens, from 92.9 percent to 95.9 percent, than jurisdictions won by Kerry, from 88.7 percent to 93.5 percent. REFERENCES McDonald, Michael P. The Turnout Rate Among Eligible Voters for U.S. States, State Politics and Policy Quarterly 2 (2002): McDonald, Michael P., and Samuel Popkin. The Myth of the Vanishing Voter. American Political Science Review 95 (2001):

39 EAC Election Day Survey Population Estimates 2004 General Election Updated: 09/19/ :02:20 Election StateLevelSummary_Master_ xls -- Population Estimates Cases = Number of Jurisdictions Reporting Subject Matter Percent Administration Estimated 2004 Citizen Cases Code Name Jurisdictions Estimated VAP Cases Citizen Vap Cases of Total VAP Cases >100% 01 Alabama 67 3,425, ,376, Alaska 1 470, , Arizona 15 4,194, ,770, Arkansas 75 2,069, ,024, California 58 26,647, ,671, Colorado 64 3,456, ,233, Connecticut 169 2,684, ,514, Delaware 3 629, , District of Columbia 1 451, , Florida 67 13,441, ,076, Georgia 159 6,534, ,159, Hawaii 5 980, , Idaho 44 1,025, , Illinois 110 9,518, ,471, Indiana 92 4,635, ,534, Iowa 99 2,274, ,221, Kansas 105 2,049, ,972, Kentucky 120 3,157, ,110, Louisiana 64 3,358, ,305, Maine 517 1,037, ,022, Maryland 24 4,200, ,940, Massachusetts 351 4,956, ,577, Michigan 83 7,616, ,369, Minnesota 87 3,872, ,736, Mississippi 82 2,139, ,118, Missouri 116 4,344, ,263, Montana , , Nebraska 93 1,316, ,272, Nevada 17 1,737, ,536, New Hampshire 242 1,000, , New Jersey 21 6,573, ,871, New Mexico 33 1,402, ,316, New York 58 14,790, ,156, North Carolina 100 6,414, ,129, North Dakota , , Ohio 88 8,680, ,532, Oklahoma 77 2,664, ,589, Oregon 36 2,766, ,594, Pennsylvania 67 9,615, ,395, Rhode Island , , South Carolina 46 3,174, ,106, South Dakota , , Tennessee 95 4,516, ,423, Texas ,263, ,443, Utah 29 1,645, ,548, Vermont , , Virginia 134 5,695, ,388, Washington 39 4,732, ,414, West Virginia 55 1,430, ,422, Wisconsin 1,910 4,188,206 1,894 4,091,525 1, , Wyoming , , American Samoa 1 66 Guam 1 72 Puerto Rico Virgin Islands 1 Total 6, ,279,989 6, ,438,494 6, ,408 Maximum 1,910 26,647,955 1,894 21,671,670 1, ,888 Average 119 4,338, ,871, Minimum 1 386, , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

40 EAC Election Day Survey Population Estimates 2004 General Election Updated: 09/19/ :02:20 Election StateLevelSummary_Master_ xls -- Population Estimates Cases = Number of Jurisdictions Reporting Subject Matter Percent Administration Estimated 2004 Citizen Cases Code Name Jurisdictions Estimated VAP Cases Citizen Vap Cases of Total VAP Cases >100% Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,612, ,583, Punch card ,552, ,456, Lever ,918, ,857, Paper 1,734 3,308,339 1,724 3,246,269 1, ,722 Optical scan 2,541 88,323,954 2,541 81,287,919 2, ,534 Electronic ,761, ,448, Multiple Systems ,803, ,559, Changed Voting Equipment Since 2000 General Election Yes 1,753 69,121,688 1,747 61,299,306 1, ,743 No 4, ,158,301 4, ,139,188 4, ,665 State Wide Voter Registration System in Place Yes 1,335 48,152,870 1,335 45,913,343 1, ,335 No 5, ,127,119 5, ,525,151 5, ,073 Election Day Registration Yes 2,823 11,509,789 2,793 11,192,644 2, ,785 No 3, ,770,200 3, ,245,850 3, ,623 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 88,988,159 1,162 78,298,249 1, ,154 In Precinct Only 4, ,866,238 4, ,935,162 4, ,327 None 1,056 7,425, ,205, No Excuse Absentee Balloting Yes 3,781 85,693,320 3,754 77,136,882 3, ,747 No 2, ,586,669 2, ,301,612 2, ,661 Early Voting Allowed Yes 1,701 99,654,623 1,701 89,386,654 1, ,701 No 4, ,625,366 4, ,051,840 4, ,707 Covered By Section 203, Language Minority Requirements Yes ,670, ,221, No 6, ,609,924 5, ,216,595 5, ,941 Covered By Section 5 of Voting Rights Act Yes ,030, ,618, No 5, ,249,505 5, ,819,764 5, ,529 Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

41 EAC Election Day Survey Population Estimates 2004 General Election Updated: 09/19/ :02:20 Election StateLevelSummary_Master_ xls -- Population Estimates Cases = Number of Jurisdictions Reporting Subject Matter Percent Administration Estimated 2004 Citizen Cases Code Name Jurisdictions Estimated VAP Cases Citizen Vap Cases of Total VAP Cases >100% Demographics Region Northeast 1,710 41,988,043 1,696 33,775,344 1, ,693 South 1,423 79,567,761 1,423 74,625,430 1, ,423 Midwest 2,902 49,563,034 2,886 45,519,951 2, ,872 West ,161, ,517, Territories 113 Urban to Rural Urban ,075, ,512, Suburban ,268, ,919, Small Towns 1,710 56,213,989 1,700 53,925,497 1, ,700 Rural 3,307 23,722,427 3,288 23,081,101 3, ,280 Not Available - Territories 113 Size of Jurisdiction (VAP) < 1,000 1, ,315 1, ,183 1, ,754 >=1,000 to <3,500 1,165 2,267,899 1,165 2,237,383 1, ,165 >=3,500 to <10,000 1,043 6,692,594 1,043 6,579,642 1, ,043 >=10,000 to <50,000 1,704 38,463,619 1,704 37,481,869 1, ,701 >=50,000 to <250, ,558, ,832, >=250,000 to <1,000, ,995, ,030, >=1,000, ,402, ,383, Not Available Race and Ethnicity Predominently NH White 6, ,258,977 6, ,752,611 6, ,249 Predominently NH Black 85 4,061, ,808, Predominently NH Native American , , Predominently Hispanic 50 12,658, ,583, Not Available , , Median Income < $25, ,079, ,873, >=$25,000 to <$30, ,220, ,840, >=$30,000 to <$35,000 1,372 28,691,481 1,372 27,644,524 1, ,370 >=$35,000 to <$40,000 1,215 50,829,468 1,215 40,245,795 1, ,211 >=$40,000 to <$45, ,717, ,603, >=$45,000 to <$50, ,092, ,951, >=$50,000 1,180 50,649,351 1,179 46,278,340 1, ,178 Not Available High School Education < 60% 126 2,401, ,070, >=60% to <70% ,653, ,226, >=70% to <80% 1,646 64,350,042 1,646 51,717,398 1, ,640 >=80% to <90% 3, ,912,781 3, ,539,905 3, ,109 >=90% ,930, ,853, Not Available , , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

42 EAC Election Day Survey Population Estimates 2004 General Election Updated: 09/19/ :02:20 Election StateLevelSummary_Master_ xls -- Population Estimates Cases = Number of Jurisdictions Reporting Subject Matter Percent Administration Estimated 2004 Citizen Cases Code Name Jurisdictions Estimated VAP Cases Citizen Vap Cases of Total VAP Cases >100% Political Battleground States in 2004 Presidential Election Yes 3,093 76,824,163 3,074 72,974,742 3, ,067 No 3, ,455,826 3, ,463,752 3, ,341 Margin of Victory in 2004 Presidential Election < 2.5% ,185, ,004, >=2.5% to < 5.0% ,781, ,849, >=5.0% to < 7.5% ,701, ,058, >=7.5% to < 10.0 % ,292, ,806, >=10.0 % 4, ,310,466 4, ,710,470 4, ,476 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 86,412,155 3,112 82,400,908 3, ,107 Bush 50% to 55% ,877, ,553, Bush < 50% 136 2,380, ,259, Kerry < 50% 150 5,883, ,524, Kerry 50% to 55% ,466, ,160, Kerry > 55% 1,161 64,245,074 1,159 49,526,130 1, ,154 Tied 25 14, , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

43 Chapter 2 Voter Registration Most jurisdictions maintain a registry of persons who are eligible to vote. To be eligible to vote, a person must be a U.S. citizen, meet a residency requirement, and have attained the age of 18 by Election Day. Persons who have been legally declared insane or mentally incompetent or who have been convicted of a felony and have not had their civil rights legally restored generally cannot vote, depending on state law. Prior to HAVA (Help America Vote Act of 2002), voter registration rolls were administered by local election officials. HAVA required states to administer voter registration, although not all states have completed the transition to a statewide voter registration system. Only 17 states had a fully functional statewide voter registration system in place for the 2004 election. To register to vote, a person must submit an application to the election authority of the local jurisdiction in which he or she resides. HAVA also requires that first-time registrants provide some form of identification. After the application has been processed, a voter registration, or confirmation, card is usually mailed to the registrant. The card assigns the registrant to a specific precinct and polling place. The registration remains active as long as the registrant lives at his or her original residence address. A person must reregister if he or he moves to a new place of residence or legally changes his or her name. The election authority will usually issue a new card if an assigned precinct or polling place is changed. To keep voter registries current, state and federal laws allow election authorities to designate a registrant as inactive if, over a period of time, the registrant has not voted in a series of elections or has not had any contact with or responded to mailings by the election jurisdiction. The National Voter Registration Act of 1993 (NVRA) prohibits election jurisdictions from removing a person from the voter registry for failure to vote [sec. 8(b)(2)] or failure to notify the registrar of a change of address within a jurisdiction [sec. 8(f)]. But the NVRA does not prohibit election jurisdictions from designating as inactive, voters who have not responded to certain address confirmation mailings [sec. 8(d)(2) mailings to confirm whether registrants continue to reside in the jurisdiction] and who have not appeared at the polls or attempted to reregister. The designation of inactive status allows election jurisdictions some administrative leeway in determining, for example, the number of signatures required for ballot access or the number of precincts, ballots, or voting machines necessary to service voters at an election. Persons may be removed from a voter registry for failure to respond to a sec. 8(d)(2) confirmation mailing if the registrant has failed to vote or appeared to vote in any election between the date of the confirmation notice and the day after the second subsequent general election for a federal office has occurred. Applicability and Coverage Question 1 of the Election Day Survey asked states to provide the number of active and inactive voters in each local election jurisdiction. But the U.S. Election Assistance Commission (EAC) did not specifically ask the states to provide the total number of registered voters at the time of the November 2004 general election. Nor did the EAC ask for the number of persons who registered to

44 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-2 September 27, 2005 vote on Election Day in the six states with EDR: Idaho, Maine, Minnesota, New Hampshire, Wisconsin, and Wyoming. 1 Election Data Services had previously collected the reported numbers of registered voters from the states for their respective jurisdictions at the November election. A comparison of the EAC survey data and Election Data Services data showed the different ways in which states report voter registration numbers. In some states, voter registration is just the number of active voters, while in others, voter registration is a combination of active and inactive voters. In four states, the determination of whether to report active and inactive voters in voter registration totals is at the discretion of individual local jurisdictions. The different ways in which states report voter registration numbers are as follows: Voter Registration Reports Include Active Voters Only (26) Alabama Georgia Mississippi South Carolina Alaska Illinois Michigan South Dakota Arizona Indiana Nevada Utah California Kentucky New Hampshire* Vermont Connecticut Maine Oregon Washington Dist. of Columbia Maryland Pennsylvania Florida Minnesota Rhode Island* Voter Registration Reports Include Active and Inactive Voters (20) Arkansas Kansas Nebraska Tennessee Colorado Louisiana New Mexico Texas Delaware Massachusetts New York Virginia Hawaii Missouri North Carolina West Virginia Idaho* Montana Oklahoma Wyoming* Reporting Active and Inactive Voters Varies by Local Jurisdiction (4) Iowa New Jersey Ohio Wisconsin* Unknown (4) American Samoa Guam Puerto Rico U.S. Virgin Islands *Voter registration reports also include voters who registered on Election Day. North Dakota does not have voter registration. Because of the differences among state voter registration reports and the fact that not every state provided uniform data on active voters, we added a special column to Table 2 called Reported Total Registration. All calculations in this study based on total voter registration use Reported Total Registration. For states that report only active voters, Reported Total Registration represents solely active voters. For states that report both active and inactive voters, Reported Total Registration is a combination of active and inactive voters. In four states Iowa, New Jersey, Ohio, and Wisconsin where local election jurisdictions decide whether to report active and inactive voters, Reported Total Registration is a combination of active and inactive voters, depending on local practice. There are also two special cases: North Dakota and Wisconsin. North Dakota has no voter registration requirement and responded Not Applicable to survey question 1. In North Dakota, anyone of voting age is allowed to vote on Election Day. As a result, in Table 2, Reported Total 1 Rhode Island is not considered an Election Day Registration (EDR) state but allows persons to register on Election Day to vote for president only.

45 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-3 September 27, 2005 Registration for North Dakota is the estimated November 2004 voting age population (VAP). Wisconsin s voter registration requirement applies only to municipalities with populations larger than 5,000. Only 337 of the state s 1,900 local election jurisdictions reported voter registration numbers on the survey. Wisconsin is building a statewide voter registration database, and some jurisdictions smaller than 5,000 provided voter registration numbers on the survey. For all other Wisconsin jurisdictions, Reported Total Registration in Table 2 is estimated November 2004 VAP. Historical Context For most eligible citizens in the United States, the first step to participate in the electoral process is to register to vote, except in North Dakota, which has no voter registration, and in the six states with EDR. For residents of all other states, the last day to register to vote prior to an election depends on state law. Table 2a provides a list of state voter registration deadlines for the November 2004 general election. Some states have different deadlines for registration by mail or in person. Some of the states with EDR have deadlines for preregistration by mail. Prior to the adoption of the NVRA in 1993, individuals had to seek out voter registration applications on their own. After NVRA, voter registration applications were more readily available at public offices, most notably motor vehicles offices. The size of voter registries increased as voter registration became easier. But voter registration rolls contain a certain amount of deadwood that is, duplicate names, erroneous or obsolete address information, and names of deceased and ineligible people still listed as active, or inactive, voters. Updating registration rolls for persons who change their places of residence is a continual challenge to registrars across the country. The 2000 census revealed that 46.7 percent of the U.S. population had moved in the previous five years. The people most likely to become deadwood on the voter registration rolls are those who moved just outside the county in which they formerly resided, which was 21.3 percent of the population. Yearly current population reports from the Census Bureau have constantly shown that about 17 percent of this nation s population moves every year. The Current Population Survey (CPS) is a monthly survey of labor statistics conducted by the Census Bureau. In November of an election year, the CPS survey includes a limited number of voting questions. Table 2b shows trends in voter registration nationally. After 1994, voter registration as a percentage of the citizen voting age population increased about seven percentage points from 71.6 percent to 78.7 percent. This corresponds to the time that states began implementing NVRA, which linked voter registration changes with driver s license agencies. Since 1996, voter registration has held steady at a little more than 82 percent of the citizen voting age population (CVAP), but now complete data on 2004 shows it increased to 86 percent. Due to holes in the data collection, the EAC survey shows just 81.2 percent of the citizen voting age population was registered in While registration as a percentage of CVAP has increased, the percentage of persons identifying themselves as a citizen of voting age and registered to vote in the Census Bureau s CPS has remained relatively constant, at 67.4 percent. Moreover, the difference between the percentage of CVAP on the CPS reported as registered and the aggregate national statistics shows that for the most recent elections, over 10 percent fewer people report being registered than the state-provided statistics indicate. What makes this difference all the more significant is that election surveys consistently find more people report voting than aggregate statistics indicate, a phenomenon

46 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-4 September 27, 2005 sometimes attributed to social desirability the desire by survey respondents to provide the socially correct answer. If people misreport that they vote when they do not, it would be expected that they would misreport registering when they have not. The most plausible explanations for the discrepancy are both deadwood, and misreporting by voters. Table 2a. State Voter Registration Deadlines for 2004 General Election 31 days before the election Nevada (mail) 30 days before the election Alaska Louisiana Pennsylvania Washington (mail) Arkansas Michigan Rhode Island Wyoming (mail) Dist. of Columbia Mississippi South Carolina Georgia Montana Tennessee Hawaii Ohio Texas 29 days before the election Arizona Florida Kentucky Virginia Colorado Indiana New Jersey 28 days before the election Illinois New Mexico 27 days before the election Missouri 25 days before the election Idaho (mail) New York North Carolina Oklahoma * 24 days before the election Idaho (in person) 21 days before the election Maryland Minnesota (mail) Nevada (in-person) Oregon (new registrants) 20 days before the election Delaware Massachusetts Utah (mail) West Virginia 18 days before the election Nebraska 15 days before the election California Kansas Washington (in person) Iowa (mail) South Dakota 14 days before the election Connecticut 13 days before the election Wisconsin (mail) 10 days before the election Alabama Iowa (in person) New Hampshire (mail) Vermont 8 days before the election Utah (in person) Vermont No deadline Maine Election Day Registration Idaho Minnesota Wisconsin Maine New Hampshire Wyoming * Registration applications may be submitted anytime, but registration cards may not be issued during the 24 days prior to an election.

47 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-5 September 27, 2005 Year Voting Age Population (VAP) Table 2b. Voter Registration Statistics, Citizen Voting Age Population (CVAP) Registered Voters Percent Registered of VAP Percent Registered of CVAP Percent Registered of CVAP (CPS) ,279, ,438, ,265, ,207, ,902, ,993, ,609, ,477, ,931, ,929, ,007, ,036, ,198, ,849, ,370, ,406, ,909, ,292, ,392, ,694, ,821, ,734, ,105, ,701, ,379, ,555, ,399, ,485, ,150, ,017, ,671, ,755, ,043, ,655, ,291, ,704, ,037, ,892, ,199, ,243, ,328, ,520, ,496, ,433, ,884, ,661, ,288, ,569, ,715, Includes all 50 states and the District of Columbia. 2 Includes 48 states and the District of Columbia (excludes North Dakota and Wisconsin). 3 Includes 47 states and the District of Columbia (excludes Iowa, North Dakota, and Wisconsin). 4 Includes 46 states and the District of Columbia (excludes Iowa, Missouri, North Dakota, and Wisconsin). 5 Includes 45 states and the District of Columbia (excludes Iowa, Kansas, Missouri, North Dakota, and Wisconsin). 6 Includes 44 states and the District of Columbia (excludes Alaska, Iowa, Kansas, Missouri, North Dakota, and Wisconsin). 7 Includes 41 states (excludes Alaska, Iowa, Kansas, Mississippi, Missouri, Nebraska, North Dakota, Wisconsin, Wyoming, and the District of Columbia). 8 Includes 40 states (excludes Alabama, Alaska, Iowa, Kansas, Mississippi, Missouri, Nebraska, North Dakota, Wisconsin, Wyoming, and the District of Columbia). Other notes: Registered voter totals from include the entire voting age population for North Dakota, which does not have voter registration, and Wisconsin, where only larger jurisdictions have voter registration. Voter registration statistics for 2004 are from the Election Day Survey. Voter registration data for 2002 and earlier is from Election Data Services Inc. Citizen voting age population (CVAP) was calculated by Dr. Michael McDonald. Voter registration rates from Current Population Survey (CPS) reports are from the U.S. Census Bureau.

48 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-6 September 27, 2005 Survey Results Table 2 presents data on active and inactive voters from question 1 on the Election Day Survey. In the table, numbers of active and inactive voters are calculated as percentages of the reported total number of registered voters as well as the VAP and thecvap. The table also provides EDR statistics for four of the six states that allow voters to register on Election Day. The column headings in Table 2 are as follows: Column Headings for Table 2. Voter Registration Col.- Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Reported Total Registration Number of active and inactive registered voters from survey questions 1a and 1b, supplemental data on Election Day registration in six states, and VAP data for North Dakota and jurisdictions in Wisconsin that do not have voter registration 5 Cases Number of jurisdictions that responded to question 1, that provided Election Day registration data, or for which VAP data was substituted for voter registration data 6 Active Registration Number of active registered voters from survey question 1a 7 Cases Number of jurisdictions that responded to question 1a, that provided Election Day registration data, or for which VAP data was substituted 8 Percent Active Registration 9 Inactive Registration Number of active registered voters (col. 6) divided by the total number of registered voters (col. 4) Number of inactive registered voters from survey question 1b 10 Cases Number of jurisdictions that responded to question 1b 11 Percent Inactive Registration 12 Percent Increase If Inactive Reg. Included 13 Election Day Registration Number of inactive registered voters (col. 9) divided by the total number of registered voters (col. 4) Number of inactive registered voters (col. 9) divided by the number of active registered voters (col. 6) Number of persons who registered to vote on Election Day (six states) 14 Cases Number of jurisdictions that provided supplemental data on Election Day registration 15 Percent Election Day Registration Number of persons who registered on Election Day (col. 13) divided by the total number of registered voters (col. 4)

49 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-7 September 27, 2005 Col.- Heading Description 16 Percent Total Registration of VAP Column Headings for Table 2 (cont.) Number of registered voters (col. 4) divided by the estimated voting age population (col. 4 of Table 1) 17 Cases > 100% Number of jurisdictions where the reported number of registered voters (col. 4) is greater than the estimated voting age population (col. 4 of Table 1) 18 Percent Active Registration of VAP Number of active registered voters (col. 6) divided by the estimated voting age population (col. 4 of Table 1) 19 Cases > 100% Number of jurisdictions where the reported number of active registered voters (col. 6) is greater than the estimated voting age population (col. 4 of Table 1) 20 Percent Total Registration of CVAP Number of registered voters (col. 4) divided by the estimated citizen voting age population (col. 6 of Table 1) 21 Cases > 100% Number of jurisdictions where the reported number of registered voters (col. 4) is greater than the estimated citizen voting age population (col. 6 of Table 1) 22 Percent Active Registration of CVAP Number of active registered voters (col. 6) divided by the estimated citizen voting age population (col. 6 of Table 1) 23 Cases > 100% Number of jurisdictions where the reported number of active registered voters (col. 6) is greater than the estimated citizen voting age population (col. 6 of Table 1) Note: VAP = Voting Age Population, CVAP = Citizen Voting Age Population.

50 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-12 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 2 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment Since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Pre-clearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 2 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary The registration data for November 2004 shows that nearly million persons were reported registered, an increase of nearly 15 million from the 2002 election and 14 million from the last presidential election in For 2004, voter registration constituted 79.5 percent of the VAP and 86.0 percent of the CVAP of the United States excluding territories, according to responses to the EAC survey. If only active voters are considered the registration base, then voter registration constituted 74.9 percent of the VAP and 81.2 percent of the CVAP. The level of inactive voters was highest in the largest jurisdictions of this nation, along with those in the West. Jurisdictions that are predominantly Hispanic and those covered by section 203 of the Voting Rights Act (language minority requirements) also have some of the highest levels of inactive voters. This may be due to decades of failure to provide voting materials in minority languages, particularly mailings to clear up registration issues that go unanswered because the potential voter does not understand the written English language. These voters would, therefore, be more likely to be moved to the inactive registration lists. Registration rates are highest in small town and rural jurisdictions, along with those that have higher education levels. For 2004, battleground states clearly had higher registration rates than nonbattleground states. The lowest registration rates can be found in predominantly Hispanic communities, but that is also a function of lower citizenship rates. States For the states that incorporate both active and inactive voters into their registration counts, the share of their rolls that is inactive varies widely. Colorado reported the largest share of inactive voters (22.5 percent). However, in 12 of the remaining 19 states, the inactive voters amount to less than 10

51 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-13 September 27, 2005 percent of the overall file. On average, inactive registrations constitute about 12 percent of a state s voter file. The lowest is in Delaware, where inactives are reportedly only 4.8 percent of the overall voter file. In the 26 states that report active voters only on voter registration rolls, and report inactive separately, we found a significantly larger number of inactive voters. The large number of inactive voters balloons the size of state voter files. On the survey, California reported another 5.6 million registered voters as inactive, which would have increased the size of the state s registration rolls by 35.4 percent. The District of Columbia voter rolls would be 44.1 percent larger if inactive voters were considered. Alaska reported registration numbers that were more than our calculated VAP and CVAP. The state did not provide information on inactive voters. Alaska s Department of Labor generates its own population estimates, and the department estimated VAP for July 1, 2004, of 461,887. As a result, the state itself reports more registered voters than their own estimated voting age population. North Dakota also hits the 100 percent mark, because the state has no voter registration and the state s voting age population is considered as registered to vote for this study. Noncitizens in North Dakota constitute about 6,000 people, and since straight VAP was used for registration, the percentage of CVAP jumps over 100 percent. When noncitizens are excluded from the base population, four states Alaska, North Dakota, Iowa, and Maine have more registered voters than CVAP. Iowa includes inactive voters in its registration counts. That totals Iowa s registered voters at over 100 percent of CVAP. However, if only active voters are considered, Iowa s registration is 94.0 percent of CVAP. Maine, on the other hand, counts only active voters as its registration base. Despite this, the number of active voters exceeds the CVAP estimate on a statewide basis, as well as in 289 of the state s 517 townships. Again, the problem of small jurisdictions is apparent in the VAP and CVAP estimates. Even Maine s planning office reports estimated VAP of just 1,010,187 for July 1, If that VAP number were aged to November, it would still be less than the registration total of 1,025,777. Table 2c presents the ranking of states by registration rates calculated against both VAP and CVAP for the state s reported registration and the state s active registration. Each state s rank changes with each of four different methods of calculating registration rates. Yet in most cases the shift is not dramatic. States near the bottom under one method tend to be near the bottom in all methods. States near the top stay near the top no matter which method is used to calculate registration rates. For the six states with EDR, we made a special effort to collect data on how many people registered on Election Day. Unfortunately, the state of Maine did not keep a separate count of these individuals in 2004, but upon the completion of a statewide voter registration system, Maine will be able to report such numbers in Not all the jurisdictions responded to our request for EDR data.

52 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-14 September 27, 2005 Table 2c. State Rankings for Registration Calculations

53 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-15 September 27, 2005 Regions Strongly influenced by California, the West is the region with the largest share of inactive voters reported on voter registration rolls, making up 26.1 percent of the region s voter file. The West also has the lowest registration rate in the nation, reporting only 68.5 percent of the VAP and 79.1 percent of the CVAP. On the other hand, the Midwest reported the highest registration rate in the nation, 89.4 percent for VAP and 93.6 for CVAP. Urban to Rural Small town and rural areas reported the lowest rates of inactive voters in voter files. Rural communities also have the highest registration rates in the nation based on voting age population (84.2 percent). However, when noncitizens are taken out of the mix, urban jurisdictions have the highest registration rates for citizens (88.7 percent). Size of Jurisdiction There is a near linear relationship between the size of the jurisdiction and the reported level of inactive voters in the voter file. Clearly, the largest jurisdictions in this nation have the largest share of inactive registered voters on their rolls. At 21.6 percent, the jurisdictions that have more than 1 million persons of voting age have more than twice as many inactive voters as the smallest jurisdictions in the nation, at 8.8 percent. The rate of registration, on the other hand, tends to be highest in the smaller jurisdictions, while the largest jurisdictions tend to have the lowest registration rates, no matter what method is used to calculate the rates. For example, nearly all (99.6 percent) persons of voting age are registered in jurisdictions with less than 1,000 population, but only 70.0 percent are registered in communities with more than 1 million persons. This order is retained when calculated as a percentage of CVAP. However, as noted earlier in this report, the smallest jurisdictions have the largest number of counties and towns that show more registered voters than the estimated VAP and CVAP. Race and Ethnicity Predominantly Hispanic communities have high levels of inactive voters on their rolls. Collectively, nearly 23.7 percent of their rolls are inactive. This compares with just 16.4 percent in predominantly African American jurisdictions and 14.9 in predominantly White communities. The predominately Native American jurisdictions in the country had the lowest levels of inactive voters, just 12.0 percent of their rolls. Predominantly Hispanic jurisdictions also have the lowest voter registration rates in the country (61.5 percent), especially when registration rates are calculated based on VAP. Rates are higher, and on par with other jurisdictions, for Hispanic areas (81.4 percent) when non-citizens are removed from the calculations. Native American jurisdictions have the highest registration rates (86.0 percent of VAP, and 87.8 percent for CVAP). Median Income Jurisdictions with a median income of $40,000 to $45,000 have the highest share of inactive voters, 19.9 percent. On the other hand, jurisdictions with the lowest median income have the lowest share of inactive voters, 9.7 percent. The lowest median income communities also have the highest registration rates. This is likely because of the rural nature of low-income jurisdictions. But the

54 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-16 September 27, 2005 authors caution against reaching too many conclusions based on this aggregate dataset. Precinct level data would allow more concrete conclusions on this subject. High School Education Jurisdictions in the second lowest category of percentage of those with a high school education reported the highest rates of inactive voters on their rolls, 18.5 percent. The lowest rate was reported by the lowest education jurisdictions, at 9.6 percent. Excepting this, the share of inactive voters tended to decrease as education increased. The higher the levels of high school education, generally the higher the reported rates of registration. The lowest education jurisdictions defied the trend slightly. This is true for both VAP and CVAP. Section 203 Language Minority Requirements Jurisdictions covered by the language minority requirements of section 203 of the Voting Rights Act appear to have nearly twice the numbers of inactive voters on their rolls (21.5 percent), compared with jurisdictions that are not covered (12.1 percent). Covered jurisdictions also have a significantly lower voter registration rate among the voting age population, 70.6 percent compared with 83.8 percent for jurisdictions not covered by section 203. The difference, however, lessens when citizenship is taken into account, 83.0 percent for covered jurisdictions versus 87.3 percent for noncovered areas. Section 5 Preclearance of Voting Procedures Section 5-covered jurisdictions have a slightly lower share of inactive voters, 15.1 versus 15.5 percent. Section 5 jurisdictions reported lower registration rates than other jurisdictions, 68.3 versus 77.2 percent for VAP and 75.6 versus 83.1 percent for CVAP. Type of Voting Equipment Jurisdictions that use lever machines and hand-counted paper ballots reported the lowest numbers of inactive voters, 9.5 and 10.4 percent, respectively. Jurisdictions using paper ballots also have the highest registration rates in the nation, 93.2 percent of VAP or 95.0 percent of CVAP. This is likely because these jurisdictions tend to be rural and in the Midwest. Jurisdictions that used multiple systems or optical scans reported the highest rates of inactive voters, 18.5 and 17.4 percent, respectively. These jurisdictions were largely in-line with the registration rates of other jurisdictions. Jurisdictions that used electronic machines reported the lowest registration rates, 75.9 percent of VAP and 82.7 percent of CVAP. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment since 2000 reported a higher percentage of inactive voters on their files, 18.9 versus However, jurisdictions that changed voting systems reported lower registration rates than other jurisdictions, 74.1 versus 82.0 for VAP and 83.5 versus 87.1 for CVAP. Statewide Voter Registration Database Jurisdictions in states with a statewide voter registration database in place for the 2004 election reported a lower percentage of inactive voters than the rest of the nation, 11.0 versus 16.5 percent.

55 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-17 September 27, 2005 Jurisdictions in states with a statewide voter registration database reported slightly lower registration rates than other jurisdictions, 78.9 versus 79.7 percent for VAP; 82.7 versus 86.9 percent for CVAP. Election Day Registration Jurisdictions that allow EDR reported lower numbers of inactive voters on their rolls than other jurisdictions, 10.5 versus 15.5 percent. In addition, EDR jurisdictions have higher registration rates than other jurisdictions, 89.7 versus 78.9 for VAP and 92.2 versus 85.6 for CVAP. Provisional Ballot Acceptance Jurisdictions in states that allowed provisional ballots to be counted if they were cast in any precinct in the home jurisdiction reported nearly twice the numbers of inactive voters compared with jurisdictions where voters could vote in their home precincts only, 20.9 versus 11.9 percent. The overall jurisdiction communities also reported lower registration rates than the in-precinct only areas, 74.0 versus 83.0 percent for VAP and 82.0 versus 88.6 percent for CVAP. No Excuse Absentee Balloting Jurisdictions that allow the more liberal practice of accepting no excuse absentee ballots reported a higher share of inactive voters, 18.6 percent versus 13.3 percent. However, no excuse absentee balloting jurisdictions reported lower rates of registration, 75.8 versus 81.9 percent of VAP and 84.2 versus 87.1 percent of CVAP. Early Voting Jurisdictions in states that allowed some form of early voting had a higher share of inactive voters, 18.2 percent versus 12.7 percent. However, these same jurisdictions tended to have lower registration rates than those in states that do not have early voting provisions, 74.5 versus 83.5 percent of VAP and 83.1 versus 88.3 percent of CVAP. Battleground States Jurisdictions in 2004 battleground states tended to have fewer inactive voters on the rolls than those in nonbattleground states, 13.8 percent versus 16.0 percent. However, the impact of the 2004 campaign can be seen in battleground states higher registration rates than nonbattleground states, 84.4 versus 76.9 percent for VAP and 88.8 versus 84.3 percent for CVAP. Presidential Margin of Victory There is no clear pattern of percentage of inactive voters by a jurisdiction s presidential margin of victory, ranging between 12.5 and 16.0 percent. There do not appear to be significant differences in the level of inactive voters or registration rates whether jurisdictions had close or large margins of victory in the 2004 presidential election, ranging between 78.2 and 85.9 percent of VAP and 84.2 and 89.4 percent of CVAP. Red versus Blue Jurisdictions Jurisdictions strongly carried by Kerry in the presidential election of 2004 reported the highest share of inactive registration, 20.0 percent, though there was no clear pattern among the remaining jurisdictions, which ranged between 12.0 and 15.3 percent. Registration rates in jurisdictions carried by Bush were similar to those jurisdictions carried by Kerry, ranging from 79.4 to 86.7 percent for

56 2004 Election Day Survey Report, Part 2 Survey Results Voter Registration, Page 2-18 September 27, 2005 Bush jurisdictions and 78.7 to 80.7 percent for Kerry jurisdictions, among VAP, and 83.2 to 90.3 percent for Bush jurisdictions and 86.3 to 88.8 percent for Kerry jurisdictions, among CVAP. REFERENCES Crocker, Royce. Congressional Research Service, Library of Congress. Voter Registration and Turnout: , CRS Report to Congress. Washington: GPO, (The CRS report utilizes data compiled by Election Data Services, Inc.) Dēmos Democracy Program. Election Day Registration Helps America Vote, Winter League of Women Voters. Voter Registration Deadlines, 21 April 2005 < Project Vote. Voter Registration Deadlines, 21 April 2005 < United States. Federal Election Commission, National Clearinghouse on Election Administration. Implementing the National Voter Registration Act of 1993: Requirements, Issues, Approaches, and Examples. Washington: GPO, 1994.

57 EAC Election Day Survey Registration 2004 General Election Registration Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :02:44 Percent Percent Percent Percent Percent Election Reported Percent Percent Increase if Percent Total Active Total Active Administration Total Active Active Inactive Inactive Inactive Reg Election Day Election Day Registration Cases Registration Cases Registr of Cases Registr of Cases Code Name Jurisdictions Registration Cases Registration Cases Registration Registration Cases Registration Included Registration Cases Registration of VAP >100% of VAP >100% Citizen VAP >100% Citizen VAP >100% 01 Alabama 67 2,597, ,597, , Alaska 1 472, , Arizona 15 2,642, ,642, , Arkansas 75 1,699, ,495, , California 58 16,646, ,646, ,674, Colorado 64 3,101, ,405, , Connecticut 169 1,831, ,831, , Delaware 3 553, , , District of Columbia 1 383, , , Florida 67 10,300, ,300, , Georgia 159 4,248, ,248, , Hawaii 5 647, , , Idaho , , , Illinois 110 7,195, ,195, ,588, Indiana 92 4,296, ,296, Iowa 99 2,226, ,080, , Kansas 105 1,695, ,582, , Kentucky 120 2,794, ,794, , Louisiana 64 2,932, ,693, , Maine 517 1,026, ,026, , Maryland 24 3,105, ,105, , Massachusetts 351 4,098, ,688, , Michigan 83 7,164, ,164, Minnesota 87 2,977, ,977, , , Mississippi 82 1,469, ,469, , Missouri 116 4,194, ,642, , Montana , , , Nebraska 93 1,160, ,160, Nevada 17 1,073, ,073, , New Hampshire , , , New Jersey 21 5,011, ,643, , New Mexico , , , New York 58 11,837, ,635, ,201, North Carolina 100 5,526, ,981, , North Dakota , Ohio 88 7,965, ,919, ,221, Oklahoma 77 2,143, ,840, , Oregon 36 2,141, ,141, , Pennsylvania 67 8,366, ,366, Rhode Island , , , South Carolina 46 2,318, ,318, , South Dakota , , , Tennessee 95 3,748, ,352, , Texas ,098, ,000, ,097, Utah 29 1,278, ,278, , Vermont , , Virginia 134 4,515, ,179, , Washington 39 3,508, ,508, , West Virginia 55 1,168, ,168, Wisconsin 1,910 4,179,774 1,894 2,439, , Wyoming , , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 2,440, ,440, Virgin Islands 1 50, , Total 6, ,265,030 6, ,877,539 4, ,695,013 3, ,287, , Maximum 1,910 16,646,555 1,894 16,646, ,674, , Average 119 3,344, ,189, , , Minimum 1 50, , , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

58 EAC Election Day Survey Registration 2004 General Election Registration Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :02:44 Percent Percent Percent Percent Percent Election Reported Percent Percent Increase if Percent Total Active Total Active Administration Total Active Active Inactive Inactive Inactive Reg Election Day Election Day Registration Cases Registration Cases Registr of Cases Registr of Cases Code Name Jurisdictions Registration Cases Registration Cases Registration Registration Cases Registration Included Registration Cases Registration of VAP >100% of VAP >100% Citizen VAP >100% Citizen VAP >100% Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,484, ,306, ,805, , Punch card ,767, ,068, ,318, , Lever ,662, ,137, ,652, , Paper 1,734 3,085,167 1,733 2,457,622 1, , , Optical scan 2,541 69,198,628 2,523 64,857,909 2, ,835,254 1, , Electronic ,068, ,987, ,386, Multiple Systems ,997, ,062, ,529, , Changed Voting Equipment Since 2000 General Election Yes 1,753 51,149,755 1,746 48,662,285 1, ,214, , No 4, ,115,275 4, ,215,254 3, ,480,027 2, , State Wide Voter Registration System in Place Yes 1,335 37,384,852 1,321 36,302,758 1, ,121,657 1, , No 5, ,880,178 5, ,574,781 3, ,573,356 1, , , Election Day Registration Yes 2,823 10,323,368 2,806 8,329,269 1, , ,287, No 3, ,941,662 3, ,548,270 3, ,317,366 2, Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,077,741 1,143 63,330,247 1, ,537, In Precinct Only 4, ,336,604 4,316 94,398,839 2, ,779,603 1, , None 1,056 8,850,685 1,053 8,148,453 1, , , No Excuse Absentee Balloting Yes 3,781 64,333,790 3,750 59,674,288 2, ,151,440 1, , , No 2, ,931,240 2, ,203,251 2, ,543,573 2, , Early Voting Allowed Yes 1,701 73,710,075 1,686 68,331,736 1, ,496,964 1, , No 4, ,554,955 4,826 97,545,803 3, ,198,049 1, ,128, , Covered By Section 203, Language Minority Requirements Yes ,756, ,954, ,555, , No 6, ,508,534 6, ,922,621 4, ,139,967 2, ,273, , Covered By Section 5 of Voting Rights Act Yes ,868, ,016, ,073, , No 5, ,396,175 5, ,860,553 4, ,621,904 2, ,286, , Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

59 EAC Election Day Survey Registration 2004 General Election Registration Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :02:44 Percent Percent Percent Percent Percent Election Reported Percent Percent Increase if Percent Total Active Total Active Administration Total Active Active Inactive Inactive Inactive Reg Election Day Election Day Registration Cases Registration Cases Registr of Cases Registr of Cases Code Name Jurisdictions Registration Cases Registration Cases Registration Registration Cases Registration Included Registration Cases Registration of VAP >100% of VAP >100% Citizen VAP >100% Citizen VAP >100% Demographics Region Northeast 1,710 34,273,670 1,709 32,199,323 1, ,316, , South 1,423 62,606,676 1,407 58,462,978 1, ,083,472 1, Midwest 2,902 44,048,138 2,879 39,961,102 1, ,007, ,034, West ,845, ,763, ,287, , Territories 113 2,490, ,490, Urban to Rural Urban ,441, ,292, ,349, , Suburban ,552, ,896, ,249, , Small Towns 1,710 44,193,768 1,690 41,652,437 1, ,265, , Rural 3,307 19,586,556 3,277 17,544,973 2, ,830,527 1, , Not Available - Territories 113 2,490, ,490, Size of Jurisdiction (VAP) < 1,000 1, ,006 1, , , , >=1,000 to <3,500 1,165 2,182,148 1,164 1,384, , , >=3,500 to <10,000 1,043 5,966,645 1,037 5,494,187 1, , , >=10,000 to <50,000 1,704 31,472,681 1,681 29,632,870 1, ,565,778 1, , >=50,000 to <250, ,992, ,811, ,148, , >=250,000 to <1,000, ,396, ,600, ,803, , >=1,000, ,867, ,174, ,788, Not Available 144 2,492, ,492, Race and Ethnicity Predominently NH White 6, ,662,585 6, ,735,903 4, ,330,247 2, ,287, , Predominently NH Black 85 3,098, ,019, , Predominently NH Native American , , , Predominently Hispanic 50 7,749, ,411, ,832, Not Available 145 2,523, ,516, , Median Income < $25, ,504, ,375, , >=$25,000 to <$30, ,917, ,142, , , >=$30,000 to <$35,000 1,372 22,970,583 1,366 21,391,385 1, ,248, , >=$35,000 to <$40,000 1,215 40,443,694 1,213 37,271, ,729, , >=$40,000 to <$45, ,780, ,581, ,686, , >=$45,000 to <$50, ,218, ,910, ,958, , >=$50,000 1,180 40,936,586 1,178 38,712, ,894, , Not Available 151 2,492, ,492, High School Education < 60% 126 1,817, ,686, , >=60% to <70% ,944, ,489, ,652, , >=70% to <80% 1,646 49,285,773 1,631 45,617,972 1, ,695, , >=80% to <90% 3,111 93,198,279 3,105 87,425,017 2, ,587,245 1, , >=90% ,495, ,141, ,589, , Not Available 151 2,523, ,516, , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

60 EAC Election Day Survey Registration 2004 General Election Registration Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :02:44 Percent Percent Percent Percent Percent Election Reported Percent Percent Increase if Percent Total Active Total Active Administration Total Active Active Inactive Inactive Inactive Reg Election Day Election Day Registration Cases Registration Cases Registr of Cases Registr of Cases Code Name Jurisdictions Registration Cases Registration Cases Registration Registration Cases Registration Included Registration Cases Registration of VAP >100% of VAP >100% Citizen VAP >100% Citizen VAP >100% Political Battleground States in 2004 Presidential Election Yes 3,093 64,166,639 3,062 59,646,074 1, ,594, ,128, No 3, ,098,391 3, ,231,465 3, ,100,135 2, , Margin of Victory in 2004 Presidential Election < 2.5% ,923, ,919, ,671, , >=2.5% to < 5.0% ,133, ,464, ,068, , >=5.0% to < 7.5% ,830, ,113, ,697, , >=7.5% to < 10.0 % 429 8,833, ,191, , , >=10.0 % 4, ,044,988 4, ,692,695 3, ,340,905 2, , Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,178,580 3,094 63,589,240 2, ,327,052 1, , Bush 50% to 55% ,682, ,709, ,256, , Bush < 50% 136 2,041, ,896, , , Kerry < 50% 150 4,850, ,674, , , Kerry 50% to 55% ,160, ,666, ,321, , Kerry > 55% 1,161 49,846,628 1,154 46,837, ,134, , Tied 25 14, , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

61 Chapter 3 Ballots Counted One would think that determining how many people turned out for an election would be a simple proposition. Not in the current state of the American election system. Different states, and even different jurisdictions in a given state, use different definitions, and, therefore, report different numbers. Some states and local jurisdictions simply report the number of people who voted as the total number of votes cast for all the candidates for the office on the ballot that received the highest number of votes. Immediately after World War II, nearly three-quarters of the country followed this practice. But as time went on and states began keeping track of vote history in voter registration files, it became apparent that slightly more people actually turned out to vote than voted for the highest office on the ballot. By 2002, thirty-nine states were reporting real turnout numbers. By 2004 several more states had either passed laws or adopted administrative procedures to report this data. The Election Day Survey, however, adds another definition to the mix by asking for the total number of ballots counted as well as the number of ballots cast. Are ballots that were rejected included in this number that is, one cast by a voter who showed up at the polls and cast what he or she thought was a valid vote? The definition section of the survey said, The number provided in response to this question should include all ballots that were counted during Election Day, absentee, early voting, or late counting for the November 2, 2004, election (e.g., paper, electronic, military, absentee, and provisional ballots. But what is being counted? We heard reports that some jurisdictions responding to this survey reported the total number of actual physical ballots or pieces of paper they counted, so that when an individual voter was provided with several ballot cards upon which to vote for different contests or measures, the number of ballots counted were two or three times the number of people who turned out. Applicability and Coverage An analysis of the 2004 data reported to the EAC showed that in 903 jurisdictions in 21 different states, including the entire state of Arkansas and most of the states of Vermont and Wisconsin, the reported number of ballots counted was identical to the number of votes cast for the office of president. The states where small numbers of jurisdictions submitted similar reports are Alabama, California, Colorado, Connecticut, Florida, Hawaii, Illinois, Iowa, Maine, Massachusetts, Missouri, Montana, New Hampshire, New Mexico, Ohio, Oklahoma, Oregon, Utah, Virginia, and Wisconsin. Data and reports from all other states clearly showed that more people participated in the election than those that just voted for president. In addition, Election Data Services maintains its own data collection of election returns and turnout measures. Comparing what we call Maximum Vote Turnout i.e., the highest of either the total voter turnout, or, where not reported, the total number of votes cast for all the candidates for the office on the ballot that received the highest number of votes to the numbers that were reported on the survey to the EAC, we found that nearly 2.4 million more people voted in the 2004 election than was reported to the EAC. Several states turned in data that was incomplete: data was missing from

62 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 2 September 27, 2005 certain local jurisdictions. For example, the state of Pennsylvania left out data for the largest counties Erie, Berks, Philadelphia, and Allegheny (Pittsburgh) along with 17 other counties. All told, 6,488 of the 6,568 jurisdictions in our database provided data on the number of ballots counted. Overall, 53 jurisdictions in the EAC database showed more ballots cast than there were registered voters in the jurisdiction. More than half of these occurred in Wisconsin, which allows Election Day registration and has nearly 2,000 townships and municipalities. Minnesota also has several jurisdictions with more than 100 percent of persons on the registration rolls turning out to vote. Like Wisconsin, Minnesota also allows Election Day registration, but the data is at the county level. When one shifts to voting age population, 78 jurisdictions showed more persons voting than the estimated voting age population, and 83 jurisdictions showed more votes cast than the estimated citizen voting age population. Most of these, however, are very small jurisdictions (notably in Maine and Wisconsin) where small numbers of people can be slightly off in the estimating process and are therefore more likely to be in contrast with other small data for the jurisdiction. For example, the estimation process may calculate 85 persons of voting age while the registration counts show there are 87 persons registered. Election administrators tend to look at turnout as measured against registered voters. These are two numbers that they know for their own jurisdiction and they can calculate for each election. On the other hand, academics and some political observers tend to calculate turnout against the base of voting age population. They do this because of the general belief that registration rolls contain varying levels of deadwood and inaccuracies and that voting age population provides a better base to use for comparative analysis. In the past several years, some academics (including the consultant on this project) have sought to modify the voting age population to take out the impact of noncitizens and other demographic groups not eligible to be part of the electorate. For the purposes of this study, we have calculated turnout using all three methods. Historical Context Traditionally, turnout in United States elections has been measured by the total number of votes cast for the highest office. In a presidential election, such as 2004, the highest office is the president. In other elections, particularly in non-presidential election years, highest office has been defined as the highest vote-getting office among U.S. Senate, governor, or the sum of all the U.S. House races in the state. Although similar sounding, the total number of ballots cast or counted is not the same as the total number of ballots cast for the highest office. Some voters, either intentionally or by error, may not record a vote for the highest office on the ballot. Yet, not all blank ballots are errors. For example, 3,688 Nevada voters, or 0.44 percent, choose None of these Candidates in the 2004 presidential race. Although that choice in Nevada is generally considered a candidate in the traditional sense of the word, Nevada s choice suggests that in states where voters do not have a similar choice, many abstain from the presidential election, but may vote for another office on the same ballot. In testimony before the EAC in May 2004, this study s author presented a historical compilation of the difference between the total number of ballots cast and the vote for highest office. It was shown

63 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 3 September 27, 2005 as percent drop-off, and is reproduced below in Table 3a. For more information on drop-off, see Chapter 7. Year Table 3a. Electoral Drop-Off Rates, Number of States Reporting Voter Turnout Actual Voter Turnout Highest Office Turnout* Drop-Off Rate ,357, ,322, % ,219,507 60,795, % ,563,022 81,059, % ,597,179 55,856, % ,638,630 69,216, % ,805,112 54,313, % ,974,912 72,629, % ,890,326 43,409, % ,081,471 56,668, % ,197,435 40,400, % ,509,636 57,113, % ,713,433 44,314, % ,797,469 54,670, % ,827,229 36,520, % ,489,395 48,377, % ,624,018 30,604, % ,582,628 41,458, % ,836,937 31,973, % ,968,112 37,389, % ,645,227 30,952, % ,724,809 36,995, % ,439,966 29,813, % ,670,435 38,076, % ,893,207 28,075, % ,935,458 33,250, % ,986,530 23,395, % ,467,386 30,985, % ,614,402 23,883, % ,121,161 27,485, % *Total number of votes cast for all the candidates for the office on the ballot that received the highest number of votes. Source: Election Data Services, Inc.

64 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 4 September 27, 2005 The Election Day Survey represents the first systematic attempt by a federal agency to collect the total number of ballots cast in a federal election. Previously, some states have reported the total number of ballots cast as a part of their election results. In post-world War II elections, 17 states reported total number of ballots cast in 1948 and 39 reported in Although the data for 2004 is not complete, the request for total number of ballots cast on the Election Day Survey has produced a greater number of voter turnout reports. Survey Results Table 3 presents data on the number of ballots counted from question 2 on the Election Day Survey. In the table, the number of ballots counted is calculated as a percentage of the reported total number of registered voters as well as the voting age population (VAP) and the citizen voting age population (CVAP). The column headings in Table 3 are as follows:

65 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 5 September 27, 2005 Column Headings for Table 3. Ballots Counted Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question Estimated VAP Est. Citizen VAP Estimated November 2004 voting age population (VAP) from col. 4 of Table 1 Estimated November 2004 citizen voting age population (CVAP) from col. 4 of Table 1 6 Cases Number of jurisdictions for which 2004 VAP and CVAP estimates were constructed 7 Total Registration Number of active and inactive registered voters, number of persons who voted on Election Day in six states, and VAP data for North Dakota and jurisdictions in Wisconsin that do not have voter registration, from col. 4 of Table 2 8 Cases Number of jurisdictions that responded to survey question 1, that provided Election Day registration data, or for which VAP data was substituted for voter registration data 9 Total Ballots Counted Number of ballots counted from survey question 2 10 Cases Number of jurisdictions that responded to question 2 11 Percent Ballots Counted of Registration Number of ballots counted (col. 9) divided by the number of registered voters (col. 7) 12 Cases Number of jurisdictions that responded to survey questions 1 and 2, that provided Election Day registration data, or for which VAP data was substituted for voter registration data 13 Cases > 100% Number of jurisdictions where the reported number of ballots counted (col. 9) is greater than the reported number of registered voters (col. 7). 14 Percent Ballots Counted of VAP Number of ballots counted (col. 9) divided by the estimated voting age population (col. 4)) 15 Cases Number of jurisdictions that responded to survey question 2 and for which 2004 VAP estimates were constructed

66 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 6 September 27, 2005 Column Headings for Table 3 (cont.) Col. Heading Description 16 Cases > 100% Number of jurisdictions where the reported number of ballots counted (col. 9) is greater than the estimated VAP (col. 4). 17 Percent Ballots Counted of Citizen VAP Number of ballots counted (col. 9) divided by the estimated citizen voting age population (col. 5) 18 Cases Number of jurisdictions that responded to survey question 2 and for which 2004 CVAP estimates were constructed 19 Cases > 100% Number of jurisdictions where the reported number of ballots counted (col. 9) is greater than the estimated CVAP (col. 5).

67 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 7 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 3 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Pre-clearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 3 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary Overall, the EAC Election Day Survey found that over million ballots were reported as counted in the 2004 general election, but not all jurisdictions reported data to the EAC. Other election studies have shown that over 123 million ballots were cast. We attribute the difference to jurisdictions not reporting total ballots counted. The EAC dataset shows that 70.4 percent of the total registered voters turned out to vote. Because states differ on whether their registration counts include inactive voters or not, we have also calculated turn-out percentages on the basis of just active registrations, which are available from all states. This has the impact of raising the percent of active registered voters that turned out to 74.6 percent. However, when voting age population is used as the denominator, only 55.8 percent of persons over 18 voted last fall. If non-citizens are excluded, the turnout rate increases to 60.7 percent of the citizen voting age population. Despite the data missing from some jurisdictions, these overall turnout rates are in line with other studies of turnout rates. States State turnout rates vary widely. The variation is widest when one studies turnout rates of registered voters because the registration numbers themselves are different based on the state s inclusion or exclusion of inactive registrations. Minnesota reported the highest turnout of registered voters (95.5 percent), while Wyoming came in second at 89.7 percent. On the opposite end of the scale, Texas reported the lowest turnout of registered voters, at just 57.3 percent. This is more likely because inactive voters were included in Texas registration figures, thereby driving up the denominator in the turnout equation. Table 3b ranks states by turnout percentages for three different methods.

68 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 8 September 27, 2005 By excluding registration from the mix and just studying the percent of the voting age population that had their ballots counted, Minnesota still remains at the top of the list. This is also the case for citizen voting age population. The northern states of Maine, Wisconsin, New Hampshire, Oregon and South Dakota round out the half dozen top turnout states for voting age population as well as citizen voting age population.

69 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 9 September 27, 2005 Table 3b Turnout Rates Sorted

70 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 10 September 27, 2005 At the bottom of the scale, Texas and Hawaii report the lowest turnout rates when compared with voting age population and citizen voting age population. Southern states dominate the lowest turnout jurisdictions. Regions Reported turnout of registered voters is highest in the West, more than likely due to the dominance of California in the region and that California excludes inactive voters from its registration counts. Turnout is lowest in the southern part of the nation. When calculating turnout rates based on voting age population, there is a reversal in the West. That region of the nation becomes the lowest in turnout of voting age population and the second lowest in turnout by citizen voting age population. The Midwest region reported the highest turnout in the nation on either basis. Urban to Rural Suburban communities in the nation reported the highest turnout rates of any population group. This was the case, for all population groups except citizen voting age, where urban areas has slightly higher turnout rate. Rural areas reported the lowest voting rates among registered and citizen voting age population. Urban areas reported the lowest voting age population turnout rate, due to the sizable non-citizen population in urban areas. When excluding non-citizen, urban areas had a nearly 10 percentage point increase in turn-out, from 53.4 for total voting age population to 62.3 for citizen voting age population. Size of Jurisdiction Reported turnout rates are generally higher in the smallest jurisdictions than in the largest jurisdictions, though the differences are slight when studying the impact of registration. The pattern is more pronounced when the voting age population and citizen voting age population is examined. For example, turn-out of voting age population was 71.4 percent in jurisdictions with less than 1,000 people, but dropped to just 47.4 percent for jurisdictions that had more than 1 million persons. Race and Ethnicity Turnout rates are highest in predominately white communities and the lowest, by a significant degree, in predominately Native American areas. This is true for both registration and citizen population based calculations. However, when overall voting age population is used, the predominately Hispanic communities had the lowest turnout rate (41.3%). Median Income Higher median income is related to the higher reported turnout rate for all methods of calculating turnout rates. High School Education Higher levels of high school education are related to higher turnout rates for all methods of calculating turnout rates.

71 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 11 September 27, 2005 Section 203 Language Minority Requirements Jurisdictions covered by Section 203 of the Voting Rights Act requiring language assistance at the polls tended to report lower turnout rates than jurisdictions not covered by the provisions. The difference is slight for registration turnout rates, largest for voting age population (over 10 percentage points), and about five percentage points for citizen voting age population. Section 5 Preclearance of Voting Procedures Jurisdictions covered by Section 5 of the Voting Rights Act reported lower voting rates than those jurisdictions not covered, for all methods of calculating turnout rates. Type of Voting Equipment Jurisdictions that used hand-counted paper ballots reported the highest turnout rates of any type of voting system for population-based turnout rates. However, when calculating turn-out as a percent of registered voters, those jurisdictions using optical scan voting equipment had the highest turnout rate of all voting systems. Jurisdictions that used lever machines had the lowest turnout rate for registration and voting age population based methods of calculating turnout rates. Surprisingly, jurisdictions that used electronic voting machines reported the lowest turnout rates when measured by citizen voting age population and the second lowest on overall voting age population. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment in the past four years reported slightly higher turnout rates among registered voters than those jurisdictions that did not change. For voting age population and citizen voting age turnout rates, jurisdictions that changed voting equipment reported slightly lower turnout rates. Statewide Voter Registration Database Jurisdictions in states with a statewide voter registration database reported slightly higher turnout rates for registration and voting age population than jurisdictions in other states. Jurisdictions with statewide voter registration databases reported a slightly lower citizen voting age population turnout rate. Election Day Registration States that allow Election Day registration reported a significantly higher turnout rate than other states for all methods of calculating turnout rates. The difference in this category was the largest of any type of election administration procedure. Provisional Ballot Acceptance Turnout rates based on voter registration are slightly higher in jurisdictions that accept provisional ballots cast anywhere in a jurisdiction than in other jurisdictions. The pattern is reversed for turnout rates calculated for voting age or citizen voting age population. No Excuse Absentee Balloting Jurisdictions that allowed no excuse absentee balloting reported a slightly higher registration and citizen voting age population turnout rate but a lower voting age population turnout rate.

72 2004 Election Day Survey Report, Part 2 Survey Results Ballots Counted, Page 3 12 September 27, 2005 Early Voting Surprisingly, early voting actually had the opposite effect from what one would have anticipated. According to the data in the EAC dataset, jurisdictions that allow early voting actually reported a lower turnout rate than other non-early-voting jurisdictions, for all methods of calculating turnout rates. Battleground States Being a battleground state clearly had a positive impact on getting out the vote. Battleground states reported higher turnout rates than other states for all methods of calculating turnout rates. Presidential Margin of Victory Presidential margin of victory within a jurisdiction was unrelated to turnout rates, regardless of which method of calculating turnout rates was used. Red versus Blue Jurisdictions Jurisdictions won by Kerry in the 2004 election tended to report a slightly higher turnout rate than those carried by President Bush, for all methods of calculating turnout rates.

73 EAC Election Day Survey Ballots Counted 2004 General Election Ballots Counted Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :03:07 Percent Percent Percent Percent Election Total Ballots Ballots Counted Ballots Ballots Administration Est. Reported Ballots Counted of Cases Of Active Cases Counted Cases Counted of Cases Code Name Jurisdictions Estimated VAP Citizen Vap Cases Registration Cases Counted Cases Registration Cases >100% Registration Cases >100% of VAP Cases >100% Citizen VAP Cases >100% 01 Alabama 67 3,425,821 3,376, ,597, ,683, Alaska 1 470, , , , Arizona 15 4,194,390 3,770, ,642, ,038, Arkansas 75 2,069,560 2,024, ,699, ,055, California 58 26,647,955 21,671, ,646, ,359, Colorado 64 3,456,263 3,233, ,101, ,148, Connecticut 169 2,684,372 2,514, ,831, ,595, Delaware 3 629, , , , District of Columbia 1 451, , , , Florida 67 13,441,568 12,076, ,300, ,639, Georgia 159 6,534,852 6,159, ,248, ,317, Hawaii 5 980, , , , Idaho 44 1,025, , , , Illinois 110 9,518,482 6,471, ,195, ,361, Indiana 92 4,635,665 4,534, ,296, ,512, Iowa 99 2,274,174 2,221, ,226, ,513, Kansas 105 2,049,512 1,972, ,695, ,199, Kentucky 120 3,157,197 3,110, ,794, ,816, Louisiana 64 3,358,452 3,305, ,932, ,956, Maine 517 1,037,050 1,022, ,026, , Maryland 24 4,200,854 3,940, ,105, ,395, Massachusetts 351 4,956,454 4,577, ,098, ,927, Michigan 83 7,616,344 7,369, ,164, ,876, Minnesota 87 3,872,349 3,736, ,977, ,842, Mississippi 82 2,139,817 2,118, ,469, ,163, Missouri 116 4,344,660 4,263, ,194, ,765, Montana , , , , Nebraska 93 1,316,475 1,272, ,160, , Nevada 17 1,737,781 1,536, ,073, , New Hampshire 242 1,000, , , , New Jersey 21 6,573,010 5,871, ,011, ,639, New Mexico 33 1,402,999 1,316, , , New York 58 14,790,540 8,156, ,837, ,448, North Carolina 100 6,414,796 6,129, ,526, ,571, North Dakota , , , , Ohio 88 8,680,792 8,532, ,965, ,730, Oklahoma 77 2,664,520 2,589, ,143, ,474, Oregon 36 2,766,936 2,594, ,141, ,851, Pennsylvania 67 9,615,172 9,395, ,366, ,006, Rhode Island , , , , South Carolina 46 3,174,262 3,106, ,318, ,626, South Dakota , , , , Tennessee 95 4,516,679 4,423, ,748, ,458, Texas ,263,861 14,443, ,098, ,507, Utah 29 1,645,366 1,548, ,278, , Vermont , , , , Virginia 134 5,695,220 5,388, ,515, ,223, Washington 39 4,732,158 4,414, ,508, ,885, West Virginia 55 1,430,254 1,422, ,168, , Wisconsin 1,910 4,188,206 4,091,525 1,888 4,179,774 1,894 3,009,491 1, , , , Wyoming , , , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 2,440, ,990, Virgin Islands 1 50, , Total 6, ,279, ,438,494 6, ,265,030 6, ,862,353 6, , , , , Maximum 1,910 26,647,955 21,671,670 1,888 16,646,555 1,894 12,359,633 1, , , , Average 119 4,338,823 3,871, ,344, ,299, Minimum 1 386, , , , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

74 EAC Election Day Survey Ballots Counted 2004 General Election Ballots Counted Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :03:07 Percent Percent Percent Percent Election Total Ballots Ballots Counted Ballots Ballots Administration Est. Reported Ballots Counted of Cases Of Active Cases Counted Cases Counted of Cases Code Name Jurisdictions Estimated VAP Citizen Vap Cases Registration Cases Counted Cases Registration Cases >100% Registration Cases >100% of VAP Cases >100% Citizen VAP Cases >100% Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,612,312 13,583, ,484, ,922, Punch card ,552,003 16,456, ,767, ,938, Lever ,918,948 19,857, ,662, ,981, Paper 1,734 3,308,339 3,246,269 1,722 3,085,167 1,733 2,172,234 1, , , , , Optical scan 2,541 88,323,954 81,287,919 2,534 69,198,628 2,523 49,661,061 2, , , , , Electronic ,761,316 48,448, ,068, ,295, Multiple Systems ,803,117 14,559, ,997, ,891, Changed Voting Equipment Since 2000 General Election Yes 1,753 69,121,688 61,299,306 1,743 51,149,755 1,746 35,479,523 1, , , , , No 4, ,158, ,139,188 4, ,115,275 4,766 86,382,830 4, , , , , State Wide Voter Registration System in Place Yes 1,335 48,152,870 45,913,343 1,335 37,384,852 1,321 27,317,939 1, , , , ,322 5 No 5, ,127, ,525,151 5, ,880,178 5,191 94,544,414 5, , , , , Election Day Registration Yes 2,823 11,509,789 11,192,644 2,785 10,323,368 2,806 8,152,145 2, , , , , No 3, ,770, ,245,850 3, ,941,662 3, ,710,208 3, , , , , Provisional Ballot Acceptance In Overall Jurisdiction 1,162 88,988,159 78,298,249 1,154 65,077,741 1,143 44,662,901 1, , , , ,115 6 In Precinct Only 4, ,866, ,935,162 4, ,336,604 4,316 69,964,775 4, , , , , None 1,056 7,425,592 7,205, ,850,685 1,053 7,234,677 1, , , No Excuse Absentee Balloting Yes 3,781 85,693,320 77,136,882 3,747 64,333,790 3,750 46,531,514 3, , , , , No 2, ,586, ,301,612 2, ,931,240 2,762 75,330,839 2, , , , , Early Voting Allowed Yes 1,701 99,654,623 89,386,654 1,701 73,710,075 1,686 50,903,807 1, , , , ,681 7 No 4, ,625, ,051,840 4, ,554,955 4,826 70,958,546 4, , , , , Covered By Section 203, Language Minority Requirements Yes ,670,065 60,221, ,756, ,287, No 6, ,609, ,216,595 5, ,508,534 6,059 87,574,692 6, , , , , Covered By Section 5 of Voting Rights Act Yes ,030,484 50,618, ,868, ,429, No 5, ,249, ,819,764 5, ,396,175 5,648 94,432,928 5, , , , , Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

75 EAC Election Day Survey Ballots Counted 2004 General Election Ballots Counted Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :03:07 Percent Percent Percent Percent Election Total Ballots Ballots Counted Ballots Ballots Administration Est. Reported Ballots Counted of Cases Of Active Cases Counted Cases Counted of Cases Code Name Jurisdictions Estimated VAP Citizen Vap Cases Registration Cases Counted Cases Registration Cases >100% Registration Cases >100% of VAP Cases >100% Citizen VAP Cases >100% Demographics Region Northeast 1,710 41,988,043 33,775,344 1,693 34,273,670 1,709 20,812,375 1, , , , , South 1,423 79,567,761 74,625,430 1,423 62,606,676 1,407 42,266,877 1, , , , ,417 1 Midwest 2,902 49,563,034 45,519,951 2,872 44,048,138 2,879 31,316,030 2, , , , , West ,161,151 43,517, ,845, ,445, Territories 113 2,490, ,021, Urban to Rural Urban ,075,044 64,512, ,441, ,675, Suburban ,268,529 55,919, ,552, ,263, Small Towns 1,710 56,213,989 53,925,497 1,700 44,193,768 1,690 30,364,561 1, , , , , Rural 3,307 23,722,427 23,081,101 3,280 19,586,556 3,277 13,536,721 3, , , , , Not Available - Territories 113 2,490, ,021, Size of Jurisdiction (VAP) < 1,000 1, , ,183 1, ,006 1, ,024 1, , , , >=1,000 to <3,500 1,165 2,267,899 2,237,383 1,165 2,182,148 1,164 1,630,543 1, , , ,162 9 >=3,500 to <10,000 1,043 6,692,594 6,579,642 1,043 5,966,645 1,037 4,256,986 1, , , , ,038 2 >=10,000 to <50,000 1,704 38,463,619 37,481,869 1,701 31,472,681 1,681 21,817,391 1, , , , ,686 >=50,000 to <250, ,558,039 57,832, ,992, ,587, >=250,000 to <1,000, ,995,785 59,030, ,396, ,485, >=1,000, ,402,590 33,383, ,867, ,427, Not Available ,492, ,022, Race and Ethnicity Predominently NH White 6, ,258, ,752,611 6, ,662,585 6, ,362,361 6, , , , , Predominently NH Black 85 4,061,404 3,808, ,098, ,117, Predominently NH Native American , , , , Predominently Hispanic 50 12,658,812 9,583, ,749, ,209, Not Available ,236 30, ,523, ,046, Median Income < $25, ,079,342 2,873, ,504, ,488, >=$25,000 to <$30, ,220,765 10,840, ,917, ,685, >=$30,000 to <$35,000 1,372 28,691,481 27,644,524 1,370 22,970,583 1,366 14,312,622 1, , , , , >=$35,000 to <$40,000 1,215 50,829,468 40,245,795 1,211 40,443,694 1,213 26,144,458 1, , , , >=$40,000 to <$45, ,717,211 44,603, ,780, ,227, >=$45,000 to <$50, ,092,115 24,951, ,218, ,037, >=$50,000 1,180 50,649,351 46,278,340 1,178 40,936,586 1,178 30,943,728 1, , , , Not Available ,492, ,022, High School Education < 60% 126 2,401,104 2,070, ,817, , >=60% to <70% ,653,549 19,226, ,944, ,083, >=70% to <80% 1,646 64,350,042 51,717,398 1,640 49,285,773 1,631 31,406,406 1, , , , ,624 8 >=80% to <90% 3, ,912, ,539,905 3,109 93,198,279 3,105 65,275,468 3, , , , , >=90% ,930,226 16,853, ,495, ,099, Not Available ,287 30, ,523, ,046, Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

76 EAC Election Day Survey Ballots Counted 2004 General Election Ballots Counted Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :03:07 Percent Percent Percent Percent Election Total Ballots Ballots Counted Ballots Ballots Administration Est. Reported Ballots Counted of Cases Of Active Cases Counted Cases Counted of Cases Code Name Jurisdictions Estimated VAP Citizen Vap Cases Registration Cases Counted Cases Registration Cases >100% Registration Cases >100% of VAP Cases >100% Citizen VAP Cases >100% Political Battleground States in 2004 Presidential Election Yes 3,093 76,824,163 72,974,742 3,067 64,166,639 3,062 43,980,255 3, , , , , No 3, ,455, ,463,752 3, ,098,391 3,450 77,882,098 3, , , , , Margin of Victory in 2004 Presidential Election < 2.5% ,185,454 18,004, ,923, ,753, >=2.5% to < 5.0% ,781,804 13,849, ,133, ,077, >=5.0% to < 7.5% ,701,432 16,058, ,830, ,931, >=7.5% to < 10.0 % ,292,117 9,806, ,833, ,126, >=10.0 % 4, ,310, ,710,470 4, ,044,988 4,463 84,945,042 4, , , , , Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 86,412,155 82,400,908 3,107 68,178,580 3,094 47,293,906 3, , , , , Bush 50% to 55% ,877,232 30,553, ,682, ,343, Bush < 50% 136 2,380,942 2,259, ,041, ,386, Kerry < 50% 150 5,883,881 5,524, ,850, ,447, Kerry 50% to 55% ,466,232 27,160, ,160, ,109, Kerry > 55% 1,161 64,245,074 49,526,130 1,154 49,846,628 1,154 33,249,808 1, , , , Tied 25 14,267 14, , , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

77 Chapter 4 Turnout Source The U.S. Election Assistance Commission (EAC) asked a number of different questions about voting in the 2004 election. There were questions on how many ballots were counted (question 2) and the number of ballots cast in polling places (question 3). There was a series of questions about various aspects of absentee balloting (questions 4, 5, and 6), along with inquiries about early voting (question 7). Finally, there were questions about provisional ballots (questions 8 and 9). Each question appears to focus on a single subject, and it is clear that some states thought about them in that manner. However, each of these questions falls under the broader subject of the different methods that people can use to vote. By considering them together, one can look at what share of the total votes come from different voting methods, or turnout source. Table 4 provides data from the Election Day Survey on ballot sources i.e., ballots cast at polling places on Election Day statewide, absentee ballots, early ballots, or provisional ballots. The first column of the Turnout Source Table shows the total number of ballots counted, as reported by the states and jurisdictions (question 2b). The number of reporting jurisdictions is reported in the next column to the right. The Sum Total Percent column is the total percentage generated by adding the percentages of each of the four voting methods identified in the table. For only 10 states, this totals 100 percent of the total ballots counted. For 23 states and territories, this calculation totals less than 100 percent, indicating either the state failed to report numbers for all four voting methods, or that ballots were left out of the count (for example, ballots separately tallied for overseas voting through the Uniformed and Overseas Citizens Absentee Voting Act [UOCAVA] survey. In 19 states and territories, this sum total percent is more than 100 percent, indicating either errors in reporting the component data, or that people reported the same ballot in two categories. For example, if absentee ballots are sent to the polling place to be counted on Election Day, it is possible that the same absentee ballot was tallied in the absentee question and the polling place question. It is also possible that in jurisdictions where the sum of percentages went well over 100 percent, that the state or local jurisdictions did not look at the original questions as components of the voting process. In all likelihood, the reported voting in precinct/polling place numbers are higher than they should be in those states and localities that total more than 100 percent. The Source Not Specified columns result from calculating the sum of the four voting methods and subtracting it from the total ballots counted. As with the Sum Total Percent column, a positive number in the Source Not Specified columns would occur when either the state failed to report numbers for any of the four categories, or they kept data separate for things like overseas voting or military voting. Negative numbers in the Source Not Specified columns occur when the four votingmethod numbers add to more than the total-ballots-counted number. The lack of data in some jurisdictions and for some questions created enormous problems in properly calculating and then analyzing the EAC dataset. It became even more difficult when

78 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-2 September 27, 2005 components of questions were expected to add to100 percent, and they did not. This chapter is particularly impacted by this problem since there is an assumption that the four methods of voting would add up to a total number of persons who voted. To calculate any percentage, one has to have both a denominator and a numerator. When creating summary calculations (like the subtotaling subject percentages), we used only those jurisdictions that reported both a numerator and denominator. However, the raw numbers shown in the tables are totals of all the available data sent to the EAC, without regard to whether there is a corresponding denominator or numerator. The Turnout Source data table contains percentage calculations for the four voting methods, and columns are entitled, Percent (voting method) of Ballots Counted. To make these calculations, both a numerator and denominator number had to be present in the dataset. In certain states and just about every subject subtotaling, not all jurisdictions had both, so where both numerator and denominator were not present, that jurisdiction s data was dropped for the purposes of making the calculation. These are the percentages that have been used for the analysis below. Because the raw numbers in the table report all data provided to the EAC, Election Data Services did a separate calculation (Level Percent) that simply takes the reported subject numerator and divides that by the Total Ballots Counted (or denominator). As a result, the further the two percentage calculations are from each other, the greater the impact of missing data. Applicability and Coverage Nearly all states have some form of absentee voting, 20 states conduct early voting, and most states offer provisional ballots, although the six states with Election Day registration and jurisdictions that do not have voter registration are exempt from provisional ballot requirements. 1 A growing number of states are opening up the absentee process by adopting a more liberal form of early voting. Statistics for early and provisional ballots counted are not reported for states that do not permit early voting or provisional balloting. (See Cross tabulation, Early Voting, and Provisional Ballot Acceptance for a list of applicable states.) Confusion concerning the distinction between questions 2 and 3 arose among some state election directors and from local jurisdictions. Question 2 requested the total number of ballots counted by all modes of voting, while question 3 requested total number of ballots cast in polling places. Some states were unresponsive to question 3, responding instead with just the total number of votes for question 2. Following phone conversations with some states, agreement was reached on how to calculate the answers to question 3 based on the responses to other questions. Confusion also arose concerning absentee ballots that were returned to polling places on Election Day or, per state law, delivered to the polling places by election administrators to be counted there. Were these ballots counted as cast on Election Day, or were they counted as absentee ballots? There 1 The six states with Election Day registration are Idaho, Maine, Minnesota, New Hampshire, Wisconsin, and Wyoming. North Dakota and jurisdictions in Wisconsin with populations under 5,000 do not have voter registration. Twenty (20) states conduct early voting: Alaska, Arizona, Arkansas, California, Colorado, Georgia, Florida, Hawaii, Idaho, Kansas, Montana, New Mexico, Nevada, North Carolina, North Dakota, Oklahoma, Tennessee, Texas, West Virginia, and Wyoming.

79 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-3 September 27, 2005 was confusion and different methods were employed by the various individuals filling out the survey, even by different local jurisdictions within the same state. Historical Context There were four modes of voting in the 2004 presidential election: voting in person at a polling place, voting by absentee ballot, voting in person early, and voting by provisional ballot. The traditional method of voting is in person in a polling place, a subject explored in detail in chapter 13. During the Civil War, the method of absentee balloting was introduced for soldiers, and was extended to civilians in the early 20 th century, a subject explored in detail in chapter 5. The Help America Vote Act established the method of provisional balloting for all federal elections starting with the November 2, 2004, election, a subject we explore in detail in chapter 6. The method of voting not treated in a separate chapter is early voting. The Election Day Survey defines early voting as any voting that occurred before November 2, 2004, for which there were no eligibility requirements. A good example of a distinction between early and absentee balloting is where, in order to obtain a ballot, voters are required to attest that they will not be present at their home on Election Day. This is a requirement, and thus would be considered an absentee, not an early vote. Synonymous with early voting is the image of polling stations in malls, libraries, and election administration offices. Most jurisdictions allow voters to cast early ballots in these easy-to-find early voting polling stations up to three weeks prior to Election Day. However, under the Election Day Survey definition of early vote, no excuse absentee balloting also may reasonably be assumed to qualify, and at least Texas and Iowa interpreted the survey item this way. For Texas and Tennessee, two states that report historical early voting numbers, the percentage of voters casting an early vote has almost tripled since In the 1994 election, Texas reported 1.3 million early votes counted, or 14.7 percent of total ballots counted. Tennessee reported 0.2 million early votes counted, or 6.0 percent of total ballots counted. By 2000, Texas reported 2.5 million early votes counted, or 20.1 percent of total ballots counted. Tennessee reported 0.7 million or 35.7 percent of total ballots counted. (Texas does not provide separate statistics for no excuse absentee balloting and early voting, while Tennessee does not have no excuse absentee balloting and thus provides separate statistics.) The EAC study shows that by 2004, Texas reported that 47.7 percent of their total votes came from early voting, while early voting was 44.9 percent of all Tennessee s total ballots cast. Survey Results Table 4 presents data on turnout sources from several questions on the Election Day Survey relating to ballots counted (question 2), ballots cast in polling places (question 3), absentee ballots counted (question 6), early voting ballots counted (question 7), and provisional ballots counted (question 8). In the table, the numbers of polling place, absentee, early voting, and provisional ballots counted are calculated as percentages of the total ballots counted. The column headings in Table 4 are as follows:

80 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-4 September 27, 2005 Table 4 Column Headings. Turnout Source Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Ballots Counted Total number of ballots counted from survey question 2 5 Cases Number of jurisdictions that responded to question 2 6 Sum Total Percent Sum of four turnout source percentages: Percent in Polling Places (col. 12), Percent Absentee of Ballots Counted (col. 18), Percent Early Voting of Ballots Counted (col. 24), and Percent Provisional of Ballots Counted (col. 30) 7 Unknown Source Unknown turnout source 8 Percent Unknown Unknown turnout source (col. 7) divided by total ballots counted (col. 4) 9 Ballots Counted in Polling Places Number of ballots cast in polling places on Election Day from survey question 3 10 Cases Number of jurisdictions that responded to question 3 11 Level Percent Number of ballots cast in polling places on Election Day (col. 9) divided by the total number of ballots counted (col. 4) 12 Percent in Polling Places Number of ballots cast in polling places on Election Day (col. 9) divided by the total number of ballots counted (col. 4), but only for jurisdictions that reported both ballots cast in polling places and total ballots counted. (Note: Raw numbers for this calculation are not in the table.) 13 Cases Number of jurisdictions that responded to questions 2 and 3 14 Cases > 100% Number of jurisdictions where the reported number of ballots cast in polling places (col. 9) is greater than the reported number of ballots counted (col. 4) 15 Absentee Ballots Counted Number of absentee ballots counted from survey question 6 16 Cases Number of jurisdictions that responded to question 6 17 Level Percent Number of absentee ballots counted (col. 15) divided by the total number of ballots counted (col. 4) 18 Percent Absentee of Ballots Counted Number of absentee ballots counted (col. 15) divided by the total number of ballots counted (col. 4), but only for jurisdictions that reported both absentee ballots counted and total ballots counted. (Note: Raw numbers for this calculation are not in the table.) 19 Cases Number of jurisdictions that responded to questions 2 and 6 20 Cases > 100% Number of jurisdictions where the reported number of absentee ballots counted (col. 15) is greater than the reported total ballots cast (col. 4).

81 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-5 September 27, 2005 Table 4 Column Headings (cont.) Col. Heading Description 21 Early Voting Ballots Counted Number of early voting ballots counted from survey question 7 22 Cases Number of jurisdictions that responded to survey question 7 23 Level Percent Number of early voting ballots counted (col. 21) divided by the total number of ballots counted (col. 4) 24 Percent Early Voting of Ballots Counted Number of early voting ballots counted (col. 21) divided by the total number of ballots counted (col. 4), but only for jurisdictions that reported both early voting ballots and total ballots counted. (Note: Raw numbers for this calculation are not in the table.) 25 Cases Number of jurisdictions that responded to questions 2 and 7 26 Cases > 100% Number of jurisdictions where the reported number of early voting ballots counted (col. 21) is greater than the reported total ballots counted (col. 4). 27 Provisional Ballots Counted Total number of provisional ballots cast from survey question 8 28 Cases Number of jurisdictions that responded to question 8 29 Level Percent Number of provisional ballots counted (col. 27) divided by the total number of ballots counted (col. 4) 30 Percent Provisional of Ballots Counted Number of provisional ballots counted (col. 27) divided by the total number of ballots counted (col. 4), but only for jurisdictions that reported both provisional ballots and total ballots counted. (Note: Raw numbers for this calculation are not in the table.) 31 Cases Number of jurisdictions that responded to questions 2 and 8 32 Number of jurisdictions where the reported number of provisional ballots counted (col. 27) is greater than the reported total ballots Cases > 100% counted (col. 4).

82 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-6 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 4 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 4 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary Of the over million ballots tallied for the EAC survey, at least 55.3 percent were cast in precincts or polling places, while nearly 12.0 percent were cast via absentee ballots. Another 8.4 percent comes from early voting ballots in jurisdictions that allow that process, and provisional ballots contributed 1.0 percent. However, because data was not provided by all states and for all types, the voting method for at least 23.3 percent (or nearly 28.4 million votes) could not be determined. A bar chart of this data is contained in Figure 4.1. In states and jurisdictions that did provide data, voting in the polling places averaged 73.3 percent of the ballots cast. Absentee ballots accounted for 13.3 percent and in the jurisdictions that allowed early voting, that method amounted to more than 23.5 percent of all votes cast. All this information totals more than 100 percent because different states tallied information differently. In addition, these percentages were generated where both numerators and denominators were available, resulting in a smaller number of jurisdictions being covered.

83 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-7 September 27, 2005 Figure 4.1 Ballots Source 2004 States Absentee voting has become a major part of some states voting processes. Washington State leads the pack with over two-thirds (68.7 percent) of its votes reported as cast absentee. California s absentee data amounted to 32.4 percent of its votes. Iowa was also very high, showing 30.3 percent of its ballots from the absentee process. Oregon is unique, in that they allow vote by mail for all persons. The total reported for that state in the Ballots Counted in Polling Place column is the votes reported to have been counted on election day, all of which were received by mail. Allowing voters to vote early is a concept that has been growing in use. One of the leading states, Texas, reports that nearly 47.7 percent of its votes were cast early. Tennessee had nearly 44.9 percent of its votes cast early, while Nevada reported nearly 41.7 percent and Arizona reported 40.8 percent as early voting. Alaska reported the highest percent of its total votes coming from accepted provisional ballots: 7.2 percent. California had the second highest, with 4.0 percent. Arizona and the District of Columbia reported that more than 3.5 percent of their ballots were provisional ballots. Regions The West Coast reported the highest rate of absentees, along with the use of provisional ballots. The South has the highest concentration of early voting, 28.8 percent.

84 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-8 September 27, 2005 Urban to Rural Provisional ballots were reportedly used most in urban areas of the country. Early voting was highest in small towns and smallest in urban jurisdictions. Absentee voting was similar across jurisdictions but highest in suburban communities. Size of Jurisdiction As a jurisdiction got larger in size it had higher levels of absentee and provisional ballots. Similar early voting levels were reported across jurisdictions, with no clear pattern to usage. Race and Ethnicity Predominantly Hispanic jurisdictions reported the highest levels of absentee and provisional ballot usage. Predominantly non-hispanic White jurisdictions reported the highest usage of early voting. Predominantly non-hispanic Black and non-hispanic Native American jurisdictions reported similar lower levels of absentee, early, and provisional voting. Median Income Jurisdictions with higher income levels tended to report higher usage of absentee balloting. The trend is reversed for early and provisional voting. High School Education Jurisdictions with higher education levels tended to report higher usage of absentee balloting. There was no discernible pattern for early voting, and some caution should be taken in interpreting the high level of early voting for the lowest education category since there were few reporting jurisdictions. Jurisdictions in the second lowest category of education reported the highest provisional balloting; however, the lowest education category reported the lowest. Section 203 Language Minority Requirements Jurisdictions covered by Section 203 of the Voting Rights Act reported higher usage of absentee, early voting, and provisional ballots. Section 5 Preclearance of Voting Procedures Jurisdictions covered by Section 5 of the Voting Rights Act reported using absentee balloting about 1.5 times less, early voting about 1.5 times more, and provisional ballots about the same as other jurisdictions. Type of Voting Equipment Jurisdictions that used optical scan voting systems reported the highest rate of absentee ballots, over three times greater than the lowest rate among jurisdictions that used lever machines. There was no discernible pattern in reported use of provisional ballots, as all jurisdictions reported about 1.0 percent, except for those that use paper ballots, which reported provisional ballots were 0.2 percent of total ballots cast. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment reported higher levels of absentee and provisional balloting than other jurisdictions. The opposite was reported for early voting jurisdictions.

85 2004 Election Day Survey Report, Part 2 Survey Results Turnout Source, Page 4-9 September 27, 2005 Statewide Voter Registration Database The existence of a statewide voter registration database in 2004 did lead to significantly lower levels of provisional ballot usage in those communities. Jurisdictions in states with a statewide voter registration database reported more than half the level of provisional balloting than other jurisdictions, 0.5 versus 1.2 percent. Jurisdictions within a state with a statewide voter registration database also reported lower levels of absentee and early voting than other jurisdictions. Election Day Registration States with Election Day registration reported much lower rates of absentee ballot usage, early voting, and provisional ballots being cast, presumably because more people registered to vote at their polling place on Election Day than other jurisdictions. Provisional Ballot Acceptance States that let provisional ballots be counted if they were cast anywhere in a jurisdiction reported nearly four times the level of provisional ballot usage compared with states that required voters to be in the correct precinct. These jurisdictions also reported twice the level of absentee voting and half the level of early voting. No Excuse Absentee Balloting Jurisdictions with no excuse absentee balloting laws reported more use of absentee ballots than other jurisdictions, 23.6 percent versus 7.1 percent. These jurisdictions, however, reported lower use of early voting, 36.6 versus 16.6 percent, but over three times the use of provisional ballots than other jurisdictions, 1.8 versus 0.5 percent. Early Voting Jurisdictions in states that allow early voting reported higher absentee balloting, 17.6 percent versus 10.0 percent, and higher provisional balloting, 1.5 versus 0.8 percent. Battleground States Jurisdictions in a battleground state reported higher levels of absentee balloting than other jurisdictions, 17.0 percent versus 11.1 percent, while levels of early and provisional balloting were similar. Presidential Margin of Victory The degree of competitiveness in the 2004 presidential election within a jurisdiction was not clearly related to the usage of absentee, early, or provisional ballots. Red versus Blue Jurisdictions Jurisdictions won by Kerry tended to report slightly higher levels of absentee ballot usage than other jurisdictions, while jurisdictions won by Bush tended to report higher levels of early voting (except for jurisdictions won by Kerry by 50 to 55 percent). There was no clear pattern among jurisdictions with regard to provisional ballot usage, except that jurisdictions won overwhelmingly by Kerry reported nearly twice the level of provisional balloting as other jurisdictions.

86 Turnout Source EAC Election Day Survey Source Cases = Number of Jurisdictions Reporting Subject Matter Turnout Source 2004 General Election Not Specified Voting in Precinct/Polling Place Absentee Voting Early Voting Provisional Ballots Updated: 09/19/ :03:33 Ballots Percent Early Percent Percent Election Total Sum Counted Percent Absentee Absentee Voting Early Voting Provisional Provisional Administration Ballots Total Unknown Percent In Polling Level In Polling Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Code Name Jurisdictions Counted Cases Percent Source Unknown Place Cases Percent Places Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% 01 Alabama 67 1,683, ,640, , , Alaska 1 314, , , , , Arizona 15 2,038, , ,129, , , Arkansas 75 1,055, , , , , , California 58 12,359, (326,085) (2.6) 7,920, ,108, , , Colorado 64 2,148, , , , , , Connecticut 169 1,595, ,452, , Delaware 3 377, , , District of Columbia 1 230, , , , , Florida 67 7,639, ,865, ,336, ,428, , Georgia 159 3,317, (387,083) (11.7) 2,642, , , , Hawaii 5 431, (13,627) (3.2) 311, , , Idaho , , , , , Illinois 110 5,361, ,147, , , Indiana 92 2,512, (511) (0.0) 2,251, , Iowa 99 1,513, (25,443) (1.7) 1,073, , , Kansas 105 1,199, , , , , Kentucky 120 1,816, ,717, , Louisiana 64 1,956, , ,801, , , Maine , (486) (0.1) 754, Maryland 24 2,395, , ,222, , , Massachusetts 351 2,927, (41,964) (1.4) 2,821, , , Michigan 83 4,876, , ,250, , , Minnesota 87 2,842, ,611, , Mississippi 82 1,163, ,163, Missouri 116 2,765, ,558, , , Montana , (76,234) (16.7) 387, , , Nebraska , , , , Nevada , , , , , New Hampshire , , , , New Jersey 21 3,639, ,409, , , New Mexico , (21,324) (6.5) 183, , , , New York 58 7,448, ,080, , , North Carolina 100 3,571, ,413, , , , North Dakota , , , , Ohio 88 5,730, ,995, , , Oklahoma 77 1,474, (85,804) (5.8) 1,324, , , Oregon * 36 1,851, , ,585, , , Pennsylvania 67 3,006, ,859, , Rhode Island , (984) (0.2) 421, , South Carolina 46 1,626, ,623, , South Dakota , , Tennessee 95 2,458, (3,169) (0.1) 1,297, , ,102, , Texas 254 7,507, (4,716) (0.1) 3,641, , ,580, , Utah , , , , , Vermont , (30) (0.0) 253, , Virginia 134 3,223, (559) (0.0) 3,001, , Washington 39 2,885, , ,982, , West Virginia , (126,060) (16.4) 740, , , , Wisconsin 1,910 3,009,491 1, ,744, ,898 1, , Wyoming , (254) (0.1) 198, , American Samoa Guam Puerto Rico 110 1,990, , ,947, , , Virgin Islands 1 31, (335) (1.1) 30, , Total 6, ,862,353 6, ,102, ,603,992 3, , ,740,215 4, , ,189,379 1, ,222 1,225,915 4, ,122 Maximum 1,910 12,359,633 1, ,080, ,920, ,108,088 1, , ,580, , Average 119 2,299, , ,572, , , , Minimum 1 31, (387,083) (16.7) 8, * Oregon does vote-by-mail. Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

87 Turnout Source EAC Election Day Survey Source Cases = Number of Jurisdictions Reporting Subject Matter Turnout Source 2004 General Election Not Specified Voting in Precinct/Polling Place Absentee Voting Early Voting Provisional Ballots Updated: 09/19/ :03:33 Ballots Percent Early Percent Percent Election Total Sum Counted Percent Absentee Absentee Voting Early Voting Provisional Provisional Administration Ballots Total Unknown Percent In Polling Level In Polling Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Code Name Jurisdictions Counted Cases Percent Source Unknown Place Cases Percent Places Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% Election Administration Voting Equipment Used in 2004 General Election None / Unknown 908 9,922, , ,286, , ,246, , Punch card ,938, ,243, ,875, ,183, , , Lever ,981, ,487, ,700, , , , Paper 1,734 2,172,234 1, , ,044,700 1, , ,782 1, , , , Optical scan 2,541 49,661,061 2,524 9,029, ,352,237 1, ,617 7,347,262 2, , ,334, ,380 1, ,714 Electronic ,295, ,231, ,384, ,269, ,173, , Multiple Systems 123 8,891, , ,959, ,548, , , Changed Voting Equipment Since 2000 General Election Yes 1,753 35,479,523 1,739 3,936, ,652, ,573,374 1, , ,826, , No 4,815 86,382,830 4,749 24,166, ,951,575 2, ,976 9,166,841 3, , ,363, ,386 3, ,309 State Wide Voter Registration System in Place Yes 1,335 27,317,939 1,322 3,961, ,051,011 1, , ,584,867 1, ,124 1,589, ,619 1, ,232 No 5,233 94,544,414 5,166 24,141, ,552,981 2, , ,155,348 3, , ,600, ,094,296 2, ,890 Election Day Registration Yes 2,823 8,152,145 2,792 2,750, ,701, ,285 1, , , No 3, ,710,208 3,696 25,352, ,902,429 2, , ,099,930 3, ,242 10,129,910 1, ,160 1,225,286 3, ,504 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 44,662,901 1,123 8,757, ,631, ,298,521 1, ,067 3,131, ,145 1, ,074 In Precinct Only 4,350 69,964,775 4,312 19,319, ,233,762 1, , ,051,666 3, , ,992, ,562 2, ,378 None 1,056 7,234,677 1,053 26, ,739,037 1, , , , , No Excuse Absentee Balloting Yes 3,781 46,531,514 3,731 3,710, ,451,170 1, ,746 9,775,880 2, , ,750, ,965 1, ,827 No 2,787 75,330,839 2,757 24,391, ,152,822 2, , ,964,335 2, ,496 5,438, ,950 2, ,295 Early Voting Allowed Yes 1,701 50,903,807 1,681 (833,530) (1.6) 32,353,422 1, , ,434,428 1, ,541 10,189,379 1, , ,108 1, ,622 No 4,867 70,958,546 4,807 28,936, ,250,570 2, , ,305,787 3, , ,807 2, ,500 Covered By Section 203, Language Minority Requirements Yes ,287, ,583, ,294, ,020, ,701, , No 6,100 87,574,692 6,038 24,519, ,309,139 3, , ,720,080 4, , ,488, ,518 3, ,685 Covered By Section 5 of Voting Rights Act Yes ,429, ,868, ,774, ,180, ,328, , No 5,688 94,432,928 5,616 24,234, ,829,587 3, , ,559,781 4, , ,860, ,510 3, ,341 * Oregon does vote-by-mail. Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

88 Turnout Source EAC Election Day Survey Source Cases = Number of Jurisdictions Reporting Subject Matter Turnout Source 2004 General Election Not Specified Voting in Precinct/Polling Place Absentee Voting Early Voting Provisional Ballots Updated: 09/19/ :03:33 Ballots Percent Early Percent Percent Election Total Sum Counted Percent Absentee Absentee Voting Early Voting Provisional Provisional Administration Ballots Total Unknown Percent In Polling Level In Polling Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Code Name Jurisdictions Counted Cases Percent Source Unknown Place Cases Percent Places Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% Demographics Region Northeast 1,710 20,812,375 1,687 9,900, ,736,138 1, , ,038,414 1, , ,813 1, ,396 South 1,423 42,266,877 1,417 5,799, ,158,163 1, , ,236,840 1, ,256 7,937, ,361 1, ,334 Midwest 2,902 31,316,030 2,871 11,584, ,057, ,241,142 2, , , , West ,445, , ,674, ,213, ,026, , Territories 113 2,021, , ,977, , , Urban to Rural Urban ,675, ,862, ,932, ,820, ,508, , Suburban ,263, ,645, ,338, ,720, ,217, , Small Towns 1,710 30,364,561 1,685 7,563, ,783,352 1, , ,573,928 1, , ,201, ,128 1, ,237 Rural 3,307 13,536,721 3,269 3,009, ,571,710 1, ,834 1,614,455 2, ,271 1,262, ,485 1, ,968 Not Available - Territories 113 2,021, , ,977, , , Size of Jurisdiction (VAP) < 1,000 1, ,024 1, , , ,901 1, , , >=1,000 to <3,500 1,165 1,630,543 1, , , , , , >=3,500 to <10,000 1,043 4,256,986 1,038 1,003, ,618, , , , >=10,000 to <50,000 1,704 21,817,391 1,689 5,165, ,888,120 1, , ,012,094 1, ,497 1,681, ,882 1, ,499 >=50,000 to <250, ,587, ,387, ,301, ,685, ,012, , >=250,000 to <1,000, ,485, ,067, ,669, ,741, ,623, , >=1,000, ,427, ,408, ,147, ,701, ,618, , Not Available 144 2,022, , ,978, , , Race and Ethnicity Predominently NH White 6, ,362,361 6,201 27,852, ,592,039 3, , ,714,675 4, , ,185,781 1, ,147 1,017,790 3, ,875 Predominently NH Black 85 2,117, , ,460, , , , Predominently NH Native American , , , , , , Predominently Hispanic 50 5,209, (220,762) (4.2) 3,480, , , , Not Available 145 2,046, , ,987, , , , Median Income < $25, ,488, , , , , , >=$25,000 to <$30, ,685, ,302, ,268, , , , >=$30,000 to <$35,000 1,372 14,312,622 1,356 3,774, ,393, ,376,230 1, ,693, , >=$35,000 to <$40,000 1,215 26,144,458 1,204 9,027, ,981, ,485, ,460, , >=$40,000 to <$45, ,227, ,698, ,288, ,508, ,331, , >=$45,000 to <$50, ,037, ,917, ,091, ,225, ,602, , >=$50,000 1,180 30,943,728 1,173 6,704, ,994, ,630, ,290, , Not Available 151 2,022, , ,978, , , High School Education < 60% , , , , , , >=60% to <70% ,083, ,120, ,195, ,483, ,069, , >=70% to <80% 1,646 31,406,406 1,630 9,462, ,020, ,947,349 1, ,258 3,703, ,695 1, ,118 >=80% to <90% 3,111 65,275,468 3,074 15,446, ,448,287 1, , ,161,642 2, , ,595, ,535 1, ,734 >=90% ,099, ,788, ,496, ,079, , , Not Available 151 2,046, , ,987, , , , * Oregon does vote-by-mail. Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

89 Turnout Source EAC Election Day Survey Source Cases = Number of Jurisdictions Reporting Subject Matter Turnout Source 2004 General Election Not Specified Voting in Precinct/Polling Place Absentee Voting Early Voting Provisional Ballots Updated: 09/19/ :03:33 Ballots Percent Early Percent Percent Election Total Sum Counted Percent Absentee Absentee Voting Early Voting Provisional Provisional Administration Ballots Total Unknown Percent In Polling Level In Polling Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Ballots Level of Ballots Cases Code Name Jurisdictions Counted Cases Percent Source Unknown Place Cases Percent Places Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% Counted Cases Percent Counted Cases >100% Political Battleground States in 2004 Presidential Election Yes 3,093 43,980,255 3,028 9,227, ,916, ,992,306 2, , ,484, , No 3,475 77,882,098 3,460 18,875, ,687,838 2, ,890 7,747,909 2, ,523 6,705, ,061 3, ,316 Margin of Victory in 2004 Presidential Election < 2.5% ,753, ,621, ,750, ,448, , , >=2.5% to < 5.0% 476 8,077, ,310, ,999, ,037, , , >=5.0% to < 7.5% 510 9,931, ,013, ,578, ,448, , , >=7.5% to < 10.0 % 429 6,126, ,551, ,452, , , , >=10.0 % 4,492 84,945,042 4,448 19,584, ,839,469 2, , ,904,864 3, , ,712,919 1, , ,002 2, ,952 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 47,293,906 3,083 10,212, ,081,362 1, ,780 5,075,995 2, , ,550, ,479 2, ,081 Bush 50% to 55% ,343, ,366, ,749, ,279, ,780, , Bush < 50% 136 1,386, , , , , , Kerry < 50% 150 3,447, , ,743, , , , Kerry 50% to 55% ,109, ,974, ,518, ,115, ,366, , Kerry > 55% 1,161 33,249,808 1,152 8,093, ,869, ,533, ,261, , Tied 25 9, , * Oregon does vote-by-mail. Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

90 Chapter 5 Absentee Ballots On its instructions to the Election Day Survey, the U.S. Election Assistance Commission (EAC) defined absentee voting as voting prior to Election Day which requires that the voter meet qualifications other than those generally required to register to vote. Such requirements might be that the voter must attest that they will be absent from their voting jurisdiction on Election Day. The Election Day Survey instructions specifically request that ballots cast by military and overseas voters not be included in responses to the survey. Statistics on military and overseas absentee ballots were collected separately through the Uniformed and Overseas Citizens Absentee Voting Act (UOCAVA) survey (which was labeled The Military and Overseas Absentee Ballot Survey ). On absentee voting, the Election Day Survey asked for the number of absentee ballots requested, the number returned, the number counted, the number not counted, and the five most common reasons for rejecting absentee ballots. Table 5 presents results from the Election Day Survey on absentee voting. Applicability and Coverage Nearly all states have some form of absentee ballots. The following 24 states provide no excuse absentee ballots: Alaska Idaho North Carolina South Dakota Arizona Iowa North Dakota Utah California Kansas Nebraska Vermont Colorado Louisiana New Mexico Washington Florida Maine Nevada Wisconsin Hawaii Montana Oklahoma Wyoming In addition, Oregon conducts elections by mail. Absentee voting is reported in Oregon for persons who request a ballot because they will be away from their normal resident address when the mail ballots are shipped to the state s voters. Those numbers are listed in Oregon s absentee column in the tables, while the vote-by-mail counts are listed in the Ballots Cast in Polling Place column. No jurisdiction among the following states reported absentee ballots requested: Alabama, Arizona (which classifies all absentee votes as early votes), Connecticut, Indiana, Kansas, Minnesota, New Hampshire, and Vermont. No jurisdiction among the following states reported absentees returned: Alabama, Arizona, Arkansas, Kansas, Minnesota, Mississippi, New Hampshire, and Vermont. No jurisdiction within the following states reported absentees counted: Arizona, Kansas, Maine, Mississippi, South Carolina, and South Dakota. No jurisdiction among 30 states reported absentees not counted. Historical Context Absentee voting was first established in the mid-1800s for soldiers during the Civil War. Northern soldiers were reported to have cast 154,045 votes in 1864 (Lee 1916). Vermont became the first state

91 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-2 September 27, 2005 to adopt absentee voting for civilians in 1896, followed by Kansas in 1901 (for rail workers only, expanded to all citizens in 1911) and Minnesota, Missouri, Nebraska, and North Dakota in 1914 (Steinbicker 1938). These early absentee balloting procedures did not operate in the same manner as today. For example, Kansas and Nebraska permitted persons traveling outside their home county to cast a vote on Election Day in the presence of an election judge in another county within the same state, which would be forwarded to the home county (Lee 1916). North Dakota was the first state to adopt absentee voting in the familiar form known to modern voters: absentee voting by mail. Like the secret ballot, the method was imported from Australia, which adopted absentee voting by mail in 1902 (Lee 1916). During and following World War I, absentee ballot laws were extended to military personnel in nearly all states and were extended to civilians in all states, except Rhode Island, Connecticut, and Kentucky by 1925 (Ray 1926). As the number of states that allowed absentee voting increased, the acceptable reasons to be permitted to cast an absentee ballot expanded. For example, by 1925, 11 states permitted sick and disabled persons the right to cast an absentee ballot (Ray 1926). Michigan was perhaps the first state to allow no excuse absentee voting by permitting absentee voting for any person necessarily absent while engaged in the pursuit of lawful business, or recreation (Election Laws of Michigan, Revision of 1936, Ch. X. Art. 3134, as quoted in Steinbicker 1938, original emphasis). Today, states allow absentee balloting under a wide range of excuses that vary among the states, including: religion, business, school, disability, and persons who live far from their polling place. Twenty-four states permit absentee voting for any reason, or no excuse. Complete historical statistics on absentee voting do not exist. A study of the 1936 election estimated that absentee ballots constituted about 2.0 percent, or 0.9 million, of all ballots counted (Steinbicker 1938). Another study estimated 4.9 percent, or 3.4 million, absentee ballots were counted in the 1960 election (Andrews 1966). These estimates are unreliable since they depend on extrapolating absentee statistics from a few states to the entire country. Compared with the historical numbers, the number of absentee ballots cast has increased in recent elections. Mitofsky International and Edison Media Research (the national exit poll organization) estimate that 16.0 percent or 16.8 million absentee ballots were counted in the 2000 presidential election. A similar percentage of 16.0 percent, or 12.5 million, were counted in the 2002 congressional election. Previous election estimates are not as reliable due to missing data for some states. At a minimum, 12.0 percent, or 8.7 million, absentee ballots were cast in the 1998 election; 11.0 percent, or 10.6 million, in the 1996 election; 8.4 percent, or 6.4 million, in the 1994 election; and 7.7 percent, or 8.1 million, in the 1992 election. Some states have kept track of absentee balloting in the past, but most states have not. In many instances, in both states and localities, absentee balloting numbers and results are combined with the results from polling place voting and reported as just a single number. Therefore, data is not available for a great deal of the country, especially on election returns. Some localities do keep separate totals on absentee voting, but keep them only at the county level, not by precinct. As more and more people vote via absentee, any kind of demographic and political analysis becomes much more difficult to conduct in those jurisdictions.

92 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-3 September 27, 2005 Survey Results Table 5 presents data on absentee ballots requested, returned, and counted from questions 4 6 on the Election Day Survey. In the table, the number of absentee ballots requested is calculated as a percentage of reported total registration, the number of absentee ballots returned is calculated as a percentage of absentee ballots requested, and the number of absentee ballots counted is calculated as a percentage of absentee ballots returned. The column headings in Table 5 are as follows: Table 5 Column Headings. Absentee Ballots Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Registration Number of active and inactive registered voters, number of persons who voted on Election Day in six states, and VAP data for North Dakota and jurisdictions in Wisconsin that do not have voter registration, from col. 4 of Table 2 5 Cases Number of jurisdictions that responded to survey question 1, that provided Election Day registration data, or for which VAP data was substituted for voter registration data 6 Total Requested Absentee Ballots Number of absentee ballots requested from survey question 4 7 Cases Number of jurisdictions that responded to question 4 8 Percent Requested of Absentee Registration Number of absentee ballots requested (col. 6) divided by the number of registered voters (col. 4) 9 Cases Number of jurisdictions that responded to survey questions 1 and 4, that provided Election Day registration data, or for which VAP data was substituted for voter registration data 10 Cases > 100% Number of jurisdictions where the reported number of absentee ballots requested (col. 6) is greater than the reported number of registered voters (col. 4) 11 Total Absentees Returned Number of absentee ballots returned from survey question 5 12 Cases Number of jurisdictions that responded to question 5 13 Percent Absentees Returned of Requested Number of absentee ballots returned (col. 11) divided by the number of absentee ballots requested (col. 6) 14 Cases Number of jurisdictions that responded to survey questions 4 and 5 15 Cases > 100% Number of jurisdictions where the reported number of absentee ballots returned (col. 11) is greater than the reported number of absentee ballots requested (col. 6)

93 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-4 September 27, 2005 Table 5 Column Headings (cont.) Col. Heading Description 16 Absentees Counted Number of absentee ballots counted from survey question 6b 17 Cases Number of jurisdictions that responded to question 6b 18 Percent Absentee Counted of Returned Number of absentee ballots counted (col. 16) divided by the number of absentee ballots returned (col. 11) 19 Cases Number of jurisdictions that responded to questions 5 and 6b 20 Cases > 100% Number of jurisdictions where the reported number of absentee ballots counted (col. 16) is greater than the number of absentee ballots returned (col. 11) 21 Total Absentees Not Counted Number of absentee ballots not counted from survey question 6c 22 Cases Number of jurisdictions that responded to question 6c 23 Percent Absentees Not Counted of Returned Number of absentee ballots not counted (col. 21) divided by the number of absentee ballots returned (col. 11) 24 Cases Number of jurisdictions that responded to questions 5 and 6c 25 Cases > 100% Number of jurisdictions where the reported number of absentee ballots not counted (col. 21) is greater than the reported number of absentee ballots returned (col. 11)

94 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-5 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 5 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 5 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary In the Election Day Survey, the EAC requested the number of absentee ballots requested, the number returned, and the number counted in each jurisdiction. Analysis is provided for four measures reported in Table 5. The total number of absentee ballots requested is calculated as a ratio to the total voter registration. The number of absentee ballots returned is calculated as a ratio to the number of absentee ballots requested. The number of absentee ballots reported counted and not counted is calculated as a ratio to the number of absentee ballots returned. States were also asked to provide the five most common reasons why the absentee ballots were rejected, although the actual numbers of ballots rejected by the reasons for rejection were not requested. The states were not asked to provide this information for their individual jurisdictions, just a statewide summary. The reasons, according to their frequency of mention by states, are as follows: Reasons for Rejecting Absentee Ballots Frequency of Mention No voter signature 11 Ballot not timely received 9 Non-matching signature 8 Elector voted early or at the polls 6 Ballot returned as undeliverable 5 Ineligible to vote 4 No ballot application on record 3 No witness signature 3 Spoiled ballot 3 Ballot missing from envelope 2 Ballot returned in unofficial envelope 2

95 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-6 September 27, 2005 Reasons for Rejecting Absentee Ballots (cont). Frequency of Mention Multiple ballots returned in one envelope 2 Elector deceased 2 Ballot replaced 1 Envelope not sealed 1 First-time voter without proper identification 1 No election official s signature on ballot 1 No residence address on return envelope 1 Since fewer jurisdictions reported absentee ballots not counted, and because this should be the reciprocal of ballots counted, the analysis below is discussed in terms of absentee ballots counted. However, because of the high rate of counting returned absentee ballots, we found little variation among jurisdictions that might provide insight into why absentee ballots were not counted beyond the reasons provided by the states. A pattern emerges in the tabulations between reported requested and returned absentee ballots. Those jurisdictions reporting a lower rate of absentee ballots requested tend to have higher rates of absentee ballots returned (correlation = -0.22). An explanation may be related to the ease of requesting an absentee ballot. Jurisdictions with no excuse absentee balloting report much higher request rates, but lower return rates, by about six percentage points, than other jurisdictions. This pattern was similar to those jurisdictions permitting early voting. We suspect jurisdictions with administrative procedures aimed to make voting more accessible have other administrative provisions (unasked on the Election Day Survey) that ease the request of absentee ballots, such as permanent absentee balloting. Where absentee ballots are more difficult to obtain, the request rates may be lower, but the return rates are higher since these voters truly desire to cast an absentee ballot. We also note that centralized management of voter registration databases increases return rates and counting of absentee ballots. Jurisdictions with statewide voter registration databases reported slightly lower request rates compared to jurisdictions in states without statewide voter registration databases, but reported a return rate almost 6 percentage points higher and a counting rate over 3 percentage points higher. There is a general pattern of absentee ballot requests according to socioeconomic status, with lower income and education jurisdictions tending to report lower rates of requesting absentee ballots than high income and education jurisdictions. However, lower socioeconomic status jurisdictions reported higher rates of return. There is also a general pattern of absentee ballot requests according to size of the jurisdiction and the urban and rural character of a jurisdiction. Small-sized and rural jurisdictions tended to report the lowest rates of absentee requests while large-sized and urban areas reported the highest rates of absentee ballot requests. Like socioeconomic status, the large population and urban areas tended to report the lowest rates of absentee ballots returned. Among Section 203 covered jurisdictions, we see a higher reported rate of requested absentee ballots, and a lower rate of return than other jurisdictions. This is consistent with the findings already

96 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-7 September 27, 2005 discussed above, that jurisdictions that report higher rates of requesting absentee ballots report fewer ballots returned than other jurisdictions, though there may be a slight amplification of this negative relationship in Section 203 jurisdictions when compared to similar tabulations, in terms of absentee ballots requested, such as jurisdictions that permit early voting or no excuse voting. This relationship may be related to the lower return rates in predominantly Hispanic and predominantly non-hispanic Native American jurisdictions, these latter jurisdictions in particular reported both a low request rate and a low return rate. States Washington reported the highest rate of absentees cast as a percentage of registration, at 64.5 percent, more than twice that of the next closest state, California. Oregon reported the lowest percentage, at 0.9 percent, but this report is deceptive since the state s balloting is completely by mail and absentees refer only to people who request a ballot because they will be away from their normal residence when the normal mail ballots are to be sent out. Next to Oregon, Louisiana had the lowest percentage of absentee ballots requested, at 1.2 percent. Twenty states and two territories reported absentee ballots requested as a percentage of registration at 5 percent or lower. The District of Columbia reported the lowest percentage of returned absentee ballots (at 72.6 percent), and five states reported return rates between 70 to 80 percent. However, states with low rates of return may count absentees returned to polling places on Election Day as votes cast within polling place, rather than as an absentee ballot. Colorado had the highest return rate of 98.0 percent, and 20 states and two territories reported return rates above 90 percent. The most consistent reporting across jurisdictions is the counting of absentees. Jurisdictions reported that most returned absentee ballots were counted. The District of Columbia reported the lowest rate of counting absentee ballots at 87.5 percent and Maryland reported the highest rate, slightly over 100 percent. Approximately 81 jurisdictions reported more absentee ballots counted than the number returned because voters turned in absentee ballots in the polling place or other locations and they were not counted as part of the returned pool of ballots. Regions The West had, by far, the highest reported rate of absentee ballots requested, due to the popularity of absentee voting within states in the region and laws that promote absentee voting. The request rate would be even higher if all of Oregon s mail-in ballots were classified as absentee. The reported request rate of absentee balloting in the West, 27.1 percent, was more than six times that of the lowest region, the Northeast, at 4.3 percent. The Midwest reported an absentee request rate of 9.0 percent and the South, 6.9 percent. The South reported the lowest rate of absentee return, 88.4 percent, followed by, in increasing order, the Northeast, 88.5 percent; the West, 86.6 percent; and the Midwest, 94.1 percent. The South reported a counting rate of returned absentee ballots of 93.7 percent; all other regions reported a counting rate around 98 percent.

97 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-8 September 27, 2005 Urban to Rural As one would expect, suburban jurisdictions reported the highest rate of requested absentee ballots (at 13.5 percent). On the other hand, urban jurisdictions reported the lowest request rate of absentee ballots, 9.5 percent. Small town jurisdictions reported the lowest rate of return, 87.3 percent, and rural areas report the highest rate of return, 92.2 percent. Urban areas reported the lowest rates of counting absentee ballots, 94.3 percent, and suburban areas reported the highest rates, 98.6 percent. Size of Jurisdiction The largest population jurisdictions, those over one million voting age population (VAP), reported a rate of requested absentee ballots of 14.8 percent, more than double that of the smallest population jurisdictions, 6.8 percent. There is no discernible pattern among jurisdictions with populations inbetween, which vary within 7.8 and 11.6 percent. On the other hand, the reported rate of return tends to decrease with population size of the jurisdiction. The largest population jurisdictions reported the lowest rate of return, 86.4 percent, while the smallest population jurisdictions reported this highest, 95.6 percent. Reported rates of counting absentee ballots were slightly over 98 percent for all jurisdictions except the largest, which reported a counting rate of 93.0 percent. Race and Ethnicity Predominantly Hispanic jurisdictions reported the highest request rate for absentee ballots, 13.6 percent, slightly more than twice the lowest reported rate in predominantly non-hispanic Black jurisdictions, at 5.7 percent. Predominantly non-hispanic White jurisdictions reported a rate, 10.9 percent, slightly lower than predominantly Hispanic jurisdictions. Predominantly non-hispanic Native American jurisdictions reported a rate, 6.1 percent, slightly higher than predominantly non- Hispanic Black jurisdictions. Predominantly Hispanic jurisdictions reported the lowest return rate, 87.5 percent, and predominantly non-hispanic Black jurisdictions reporting the highest return rate, 90.7 percent. Predominantly non-hispanic White jurisdictions reported the lowest counting rate, 96.8 percent, and predominantly non-hispanic Black jurisdictions reported the highest counting rate, 99.7 percent. Median Income Reported rates of absentee ballots requested tend to increase with jurisdiction median income, though the rate slightly drops off at the highest income level. The reported request rate for the highest income jurisdictions, 13.8 percent, is about three times greater than the lowest income jurisdictions, 4.7 percent. Reported rates of return tend higher for lower income jurisdictions than for higher income jurisdictions. The lowest income jurisdictions reported the highest rate of return, 92.2 percent,

98 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-9 September 27, 2005 slightly more than five percentage points greater than jurisdictions with median income $45,000 50,000, at 87.4 percent. Reported rates of counting absentee ballots tended not to vary in a discernible pattern with median income. The lowest rate of counting was 92.8 percent for jurisdiction median income $40,000 45,000, and the highest rate was 99.3 percent for jurisdictions with median income $45,000 $50,000. High School Education Reported rates of absentee ballots requested tend to increase with education levels, except for the second lowest level of education, which reported a rate much higher than the trend, 12.5 percent. Jurisdictions with the lowest level of education reported a request rate of 3.4 percent and those at the highest level reported 17.8 percent. Reported rates of absentee ballots returned tend to increase with education levels, except that the highest rate of return was reported by jurisdictions with the lowest rates of education, 97.2 percent. This surprising result may be related to the small request rate within these jurisdictions. The second lowest education category reported a return rate of 96.0 percent and the highest reported 90.2 percent. Reported rates of counting absentee ballots tend not to vary greatly with education, between 96.3 and 98.6 percent. Section 203 Language Minority Requirements Jurisdictions covered by Section 203 reported almost twice the absentee request rate than other jurisdictions, 15.9 versus 8.6 percent. Jurisdictions covered by Section 203 reported a return rate five percentage points lower than other jurisdictions, 85.9 versus 90.9 percent. Jurisdictions covered by Section 203 reported a counting rate four percentage points lower than other jurisdictions, 94.6 versus 98.6 percent. Section 5 Preclearance of Voting Procedures Jurisdictions covered by Section 5 reported a lower absentee request rate than other jurisdictions, 7.3 versus 11.8 percent. Jurisdictions covered by Section 5 reported a return rate slightly higher than other jurisdictions, 90.4 versus 88.3 percent. Jurisdictions covered by Section 5 reported a counting rate slightly higher than other jurisdictions, 98.4 versus 96.6 percent. Type of Voting Equipment Optical scan voting equipment jurisdictions reported the highest absentee ballot request rate, 14.5 percent, more than four times higher than the lowest reported rate for lever jurisdictions, 3.2 percent. Optimal scan jurisdictions were followed by, in descending order: multiple-systems, electronic, punch-card, and paper-equipment jurisdictions. Lever voting-equipment jurisdictions reported the lowest rate of absentee ballot return, 85.6 percent. Paper jurisdictions had the highest rate, 95.2 percent. Paper jurisdictions were followed by, in descending order: multiple-systems, electronic, optical-scan, punch-card, and lever jurisdictions.

99 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-10 September 27, 2005 Reported rates of counting absentee ballots were above 97 percent for all jurisdictions except electronic, at 93.5 percent. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment reported higher rates of requesting absentee ballots than other jurisdictions, 14.2 versus 9.3 percent. Jurisdictions that changed voting equipment reported slightly lower rates of returned absentee ballots than other jurisdictions, 88.4 versus 88.8 percent. Jurisdictions that changed voting equipment reported lower rates of counting absentee ballots than other jurisdictions, 95.1 versus 98.0 percent. Statewide Voter Registration Database Jurisdictions with statewide voter registration databases reported lower rates of requesting absentee ballots than other jurisdictions, 8.7 versus 11.3 percent. Jurisdictions with statewide voter registration databases reported higher rates of returned absentee ballots than other jurisdictions, 93.4 versus 87.8 percent. Jurisdictions with statewide voter registration databases reported higher rates of counting absentee ballots than other jurisdictions, 99.0 versus 96.4 percent. Election Day Registration Jurisdictions with Election Day registration reported slightly lower rates of requesting absentee ballots than other jurisdictions, 10.2 versus 10.8 percent. Jurisdictions with Election Day registration reported higher rates of returned absentee ballots than other jurisdictions, 94.7 versus 88.4 percent. Jurisdictions with Election Day registration reported slightly lower rates of counting absentee ballots than other jurisdictions, 96.6 versus 96.9 percent. Provisional Ballot Acceptance Jurisdictions employing within-jurisdiction provisional ballot acceptance reported a higher rate of requesting absentee ballots than jurisdictions employing within-precinct provisional ballot acceptance, 16.1 versus 7.6 percent. Jurisdictions employing within-jurisdiction provisional ballot acceptance experienced lower rates of returned absentee ballots than jurisdictions employing withinprecinct provisional ballot acceptance, 87.4 versus 90.0 percent. Jurisdictions employing withinjurisdiction provisional ballot acceptance experienced higher rates of counting absentee ballots than jurisdictions employing within-precinct provisional ballot acceptance, 98.0 versus 95.4 percent. No Excuse Absentee Balloting Jurisdictions with no excuse absentee balloting reported almost four times the rate of requesting absentee ballots than other jurisdictions, 20.1 versus 5.1 percent. Jurisdictions with no excuse absentee balloting reported lower rates of returned absentee ballots than other jurisdictions, 87.1 versus 92.3 percent. Jurisdictions with no excuse absentee balloting reported lower rates of counted absentee ballots than other jurisdictions, 96.1 versus 98.5 percent. Early Voting Jurisdictions with early voting reported almost twice the rate of requesting absentee ballots than other jurisdictions, 14.6 versus 7.8 percent. Jurisdictions with early voting reported lower rates of returned absentee ballots than other jurisdictions, 87.5 versus 90.3 percent. Jurisdictions with early

100 2004 Election Day Survey Report, Part 2 Survey Results Absentee Ballots, Page 5-11 September 27, 2005 voting reported lower rates of counting absentee ballots than other jurisdictions, 95.4 versus 98.9 percent. Battleground States Jurisdictions in battleground states reported a higher rate of requesting absentee ballots than other jurisdictions, 14.2 versus 8.9 percent. Jurisdictions in battleground states reported slightly higher rates of returned absentee ballots than other jurisdictions, 89.3 versus 88.1 percent. Jurisdictions in battleground states reported slightly lower rates of counting absentee ballots than other jurisdictions, 96.0 versus 97.6 percent. Presidential Margin of Victory The reported absentee ballot request rate is similar among jurisdictions according to the presidential margin of victory, ranging between 11.1 and 13.1 percent. The reported absentee ballots return rate is similar among jurisdictions according to the presidential margin of victory, ranging between 88.0 and 90.3 percent. The reported absentee ballot counting rate is similar among jurisdictions according to the presidential margin of victory, ranging between 96.1 and 98.8 percent. It should be noted that the lowest rate of counting absentee ballots was from the least competitive jurisdictions. Red versus Blue Jurisdictions Jurisdictions won by Kerry tended to have slightly higher rates of requesting absentee ballots than jurisdictions won by Bush, averaging about 13 percent among jurisdictions won by Kerry and 10 percent for those won by Bush. The reported rate of returning absentee ballots was similar across jurisdictions with regard to the presidential vote within the jurisdiction, ranging between 88.5 and 90.4 percent. The reported rate of counting absentee ballots was similar across jurisdictions with regard to the presidential vote within the jurisdiction, above 97.7 percent for all jurisdictions except those won overwhelmingly by Kerry, which reported a counting rate of 93.9 percent. REFERENCES Andrews, William G American Voting Participation. The Western Political Quarterly 19(4): Lee, Duncan Campbell [article title?] Journal of the Society of Comparative Legislation 16(2): Steinbicker, Paul G Absentee Voting in the United States. The American Political Science Review 32(5): Ray, P. Orman Absent-voting Legislation, The American Political Science Review 20(2):

101 EAC Election Day Survey Absentee Ballots 2004 General Election Absentee Cases = Number of Jurisdictions Reporting Subject Matter Requested Absentees Returned Absentees Counted Absentees Not Counted Absentees Updated: 09/19/ :03:57 Total Percent Percent Percent Percent Election Requested Requested Total Absentees Absentee Total Calculated Absentees Administration Total Absentee Absentee of Cases Absentees Returned of Cases Absentees Counted of Cases Absentees Absentees Not Counted Cases Code Name Jurisdictions Registration Cases Ballots Cases Registration Cases >100% Returned Cases Requested Cases >100% Counted Cases Returned Cases >100% Not Counted Cases Not Counted of Returned Cases >100% 01 Alabama 67 2,597, , Alaska 1 472, , , , , Arizona 15 2,642, Arkansas 75 1,699, , , California 58 16,646, ,866, ,181, ,108, , , Colorado 64 3,101, , , , , Connecticut 169 1,831, , , , Delaware 3 553, , , , District of Columbia 1 383, , , , Florida 67 10,300, ,820, ,526, ,336, , , Georgia 159 4,248, , , , , , Hawaii 5 647, , , , Idaho , , , , Illinois 110 7,195, , , , , Indiana 92 4,296, , , Iowa 99 2,226, , , , , , Kansas 105 1,695, Kentucky 120 2,794, , , , , , Louisiana 64 2,932, , , , Maine 517 1,026, , , Maryland 24 3,105, , , , , , Massachusetts 351 4,098, , , , , Michigan 83 7,164, , , , , , Minnesota 87 2,977, , Mississippi 82 1,469, , Missouri 116 4,194, , , , , Montana , , , , , , Nebraska 93 1,160, , , , Nevada 17 1,073, , , , , New Hampshire , , New Jersey 21 5,011, , , , , , New Mexico , , , , , New York 58 11,837, , , , , , North Carolina 100 5,526, , , , , , North Dakota , , , , Ohio 88 7,965, , , , , Oklahoma 77 2,143, , , , , , Oregon 36 2,141, , , , Pennsylvania 67 8,366, , , , , , Rhode Island , , , , South Carolina 46 2,318, , , South Dakota , , , Tennessee 95 3,748, , , , , Texas ,098, , , , , , Utah 29 1,278, , , , , Vermont , , Virginia 134 4,515, , , , , Washington 39 3,508, ,210, ,916, ,982, , West Virginia 55 1,168, , , , Wisconsin 1,910 4,179,774 1, ,719 1, , ,366 1, ,898 1, , Wyoming , , , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 2,440, , , , Virgin Islands 1 50, , , , Total ,265,030 6,512 16,870,660 4, , ,851,332 4, ,740,215 4, ,990 1, , Maximum 1,910 16,646,555 1,894 4,866,605 1, , ,181,809 1, ,108,088 1, , , Average 119 3,344, , , , , , Minimum 1 50, , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

102 EAC Election Day Survey Absentee Ballots 2004 General Election Absentee Cases = Number of Jurisdictions Reporting Subject Matter Requested Absentees Returned Absentees Counted Absentees Not Counted Absentees Updated: 09/19/ :03:57 Total Percent Percent Percent Percent Election Requested Requested Total Absentees Absentee Total Calculated Absentees Administration Total Absentee Absentee of Cases Absentees Returned of Cases Absentees Counted of Cases Absentees Absentees Not Counted Cases Code Name Jurisdictions Registration Cases Ballots Cases Registration Cases >100% Returned Cases Requested Cases >100% Counted Cases Returned Cases >100% Not Counted Cases Not Counted of Returned Cases >100% Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,484, , , , , , Punch card ,767, ,334, ,193, ,183, , , Lever ,662, , , , , , Paper 1,734 3,085,167 1, ,260 1, , ,285 1, ,782 1, , , Optical scan 2,541 69,198,628 2,523 8,310,070 2, , ,231,586 1, ,347,262 2, , , Electronic ,068, ,962, ,588, ,269, , , Multiple Systems ,997, ,679, ,390, ,548, , , Changed Voting Equipment Since 2000 General Election Yes 1,753 51,149,755 1,746 6,504,908 1, , ,738,187 1, ,573,374 1, , , No 4, ,115,275 4,766 10,365,752 3, , ,113,145 3, ,166,841 3, ,458 1,005-53, State Wide Voter Registration System in Place Yes 1,335 37,384,852 1,321 2,651, ,615,511 1, ,584,867 1, , , No 5, ,880,178 5,191 14,219,644 3, , ,235,821 3, ,155,348 3, , , Election Day Registration Yes 2,823 10,323,368 2, ,096 1, , ,973 1, ,285 1, , No 3, ,941,662 3,706 16,305,564 2, ,896 14,311,359 2, ,099,930 3, ,663 1, , Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,077,741 1,143 9,632, ,999, ,298,521 1, , , In Precinct Only 4, ,336,604 4,316 6,972,065 3, , ,591,782 3, ,051,666 3, , , None 1,056 8,850,685 1, , , , , No Excuse Absentee Balloting Yes 3,781 64,333,790 3,750 11,763,686 2, , ,948,196 2, ,775,880 2, , , No 2, ,931,240 2,762 5,106,974 2, ,003 4,903,136 2, ,964,335 2, , , Early Voting Allowed Yes 1,701 73,710,075 1,686 9,923,747 1, ,287 8,352,808 1, ,434,428 1, , , No 4, ,554,955 4,826 6,946,913 3, , ,498,524 3, ,305,787 3, , , Covered By Section 203, Language Minority Requirements Yes ,756, ,377, ,278, ,020, , , No 6, ,508,534 6,059 9,493,322 4, , ,573,241 4, ,720,080 4, ,551 1, , Covered By Section 5 of Voting Rights Act Yes ,868, ,622, ,270, ,180, , , No 5, ,396,175 5,648 14,248,558 3, , ,580,739 4, ,559,781 4, , , Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

103 EAC Election Day Survey Absentee Ballots 2004 General Election Absentee Cases = Number of Jurisdictions Reporting Subject Matter Requested Absentees Returned Absentees Counted Absentees Not Counted Absentees Updated: 09/19/ :03:57 Total Percent Percent Percent Percent Election Requested Requested Total Absentees Absentee Total Calculated Absentees Administration Total Absentee Absentee of Cases Absentees Returned of Cases Absentees Counted of Cases Absentees Absentees Not Counted Cases Code Name Jurisdictions Registration Cases Ballots Cases Registration Cases >100% Returned Cases Requested Cases >100% Counted Cases Returned Cases >100% Not Counted Cases Not Counted of Returned Cases >100% Demographics Region Northeast 1,710 34,273,670 1,709 1,212, ,132,816 1, ,038,414 1, , , South 1,423 62,606,676 1,407 4,132,669 1, ,308 3,525,381 1, ,236,840 1, , , Midwest 2,902 44,048,138 2,879 3,149,905 1, , ,233,595 2, ,241,142 2, , , West ,845, ,372, ,948, ,213, , , Territories 113 2,490, , , , Urban to Rural Urban ,441, ,670, ,033, ,820, , , Suburban ,552, ,265, ,593, ,720, , , Small Towns 1,710 44,193,768 1,690 4,141,911 1, , ,651,437 1, ,573,928 1, , , Rural 3,307 19,586,556 3,277 1,789,608 2, ,354 1,562,217 2, ,614,455 2, , , Not Available - Territories 113 2,490, , , , Size of Jurisdiction (VAP) < 1,000 1, ,006 1,757 38,191 1, , ,406 1, ,901 1, >=1,000 to <3,500 1,165 2,182,148 1, , , , , >=3,500 to <10,000 1,043 5,966,645 1, , , , , , >=10,000 to <50,000 1,704 31,472,681 1,681 2,066,132 1, ,358 2,004,802 1, ,012,094 1, , , >=50,000 to <250, ,992, ,149, ,754, ,685, , , >=250,000 to <1,000, ,396, ,400, ,466, ,741, , , >=1,000, ,867, ,627, ,003, ,701, , , Not Available 144 2,492, , , , Race and Ethnicity Predominently NH White 6, ,662,585 6,234 15,612,881 4, , ,793,684 4, ,714,675 4, ,485 1,236 79, Predominently NH Black 85 3,098, , , , , , Predominently NH Native American , , , , , Predominently Hispanic 50 7,749, ,005, , , , , Not Available 145 2,523, , , , Median Income < $25, ,504, , , , , , >=$25,000 to <$30, ,917, , , , , , >=$30,000 to <$35,000 1,372 22,970,583 1,366 1,574,055 1, ,049 1,401,035 1, ,376,230 1, , , >=$35,000 to <$40,000 1,215 40,443,694 1,213 2,936, ,500, ,485, , , >=$40,000 to <$45, ,780, ,161, ,813, ,508, , , >=$45,000 to <$50, ,218, ,475, ,181, ,225, , , >=$50,000 1,180 40,936,586 1,178 5,057, ,427, ,630, , , Not Available 151 2,492, , , , High School Education < 60% 126 1,817, , , , , , >=60% to <70% ,944, ,769, ,458, ,483, , , >=70% to <80% 1,646 49,285,773 1,631 3,540,501 1, ,280 3,036,101 1, ,947,349 1, , , >=80% to <90% 3,111 93,198,279 3,105 9,361,286 2, , ,498,560 2, ,161,642 2, , , >=90% ,495, ,132, ,787, ,079, , , Not Available 151 2,523, , , , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

104 EAC Election Day Survey Absentee Ballots 2004 General Election Absentee Cases = Number of Jurisdictions Reporting Subject Matter Requested Absentees Returned Absentees Counted Absentees Not Counted Absentees Updated: 09/19/ :03:57 Total Percent Percent Percent Percent Election Requested Requested Total Absentees Absentee Total Calculated Absentees Administration Total Absentee Absentee of Cases Absentees Returned of Cases Absentees Counted of Cases Absentees Absentees Not Counted Cases Code Name Jurisdictions Registration Cases Ballots Cases Registration Cases >100% Returned Cases Requested Cases >100% Counted Cases Returned Cases >100% Not Counted Cases Not Counted of Returned Cases >100% Political Battleground States in 2004 Presidential Election Yes 3,093 64,166,639 3,062 7,822,484 2, , ,568,964 1, ,992,306 2, , , No 3, ,098,391 3,450 9,048,176 2, ,676 8,282,368 2, ,747,909 2, , , Margin of Victory in 2004 Presidential Election < 2.5% ,923, ,580, ,479, ,448, , , >=2.5% to < 5.0% ,133, ,146, , ,037, , , >=5.0% to < 7.5% ,830, ,703, ,338, ,448, , , >=7.5% to < 10.0 % 429 8,833, , , , , , >=10.0 % 4, ,044,988 4,463 11,472,993 3, , ,206,247 3, ,904,864 3, ,825 1, , Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,178,580 3,094 5,812,982 2, , ,264,857 2, ,075,995 2, , , Bush 50% to 55% ,682, ,570, ,071, ,279, , , Bush < 50% 136 2,041, , , , , , Kerry < 50% 150 4,850, , , , , , Kerry 50% to 55% ,160, ,337, ,115, ,115, , Kerry > 55% 1,161 49,846,628 1,154 5,326, ,641, ,533, , , Tied 25 14, Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

105 Chapter 6 Provisional Ballots Table 6 presents data from the Election Day Survey on provisional ballots. Under the Help America Vote Act of 2002 (HAVA), if an individual appears at the polls on Election Day to cast a ballot but is not listed on the voter registration rolls, that individual is permitted to cast a provisional ballot. And if the individual is later determined to be eligible to vote, the provisional ballot is counted as a vote. The Election Day Survey asked for the number of provisional ballots cast, the number counted, and the five most common reasons for rejecting provisional ballots. Section 302(a) of HAVA establishes the process of provisional balloting: If an individual declares that such individual is a registered voter in the jurisdiction in which the individual desires to vote and that the individual is eligible to vote in an election for Federal office, but the name of the individual does not appear on the official list of eligible voters for the polling place or an election official asserts that the individual is not eligible to vote, such individual shall be permitted to cast a provisional ballot as follows. The applicability of provisional balloting covers individuals who appear at the polls on Election Day to cast a ballot but are not listed on the voter registration rolls; in some states, first-time voters who cannot provide identification, as required under HAVA; and in some states, voters who were challenged at the poll. Election administrators are required to notify individuals of their opportunity to cast a provisional ballot. While all individuals may cast a provisional ballot, the states differed in their interpretation of the phrase registered voter in the jurisdiction in which the individual desires to vote as to what defines a valid provisional ballot: is the jurisdiction an individual s voting precinct, county, or some other jurisdiction? The ambiguity in the HAVA language resulted in controversy in the 2004 election and lawsuits seeking to expand the definition of jurisdiction when counting provisional ballots. In 2004, as detailed below, in 18 states provisional ballots were eligible to be counted if cast outside the individual s home precinct. In 25 states, provisional ballots were disqualified if cast outside the individual s home precinct. Seven states with Election Day registration were exempt from the HAVA provision, but three of these adopted provisional ballots for some classes of individuals seeking to vote. If the individual is later determined to be eligible to vote, the provisional ballot is counted as a vote. A 2004 survey of 35 state election administrators by the National Association of Secretaries of State (NASS) revealed that for the 2004 election states begin verification procedures as early as Election Day and as late as one week after the election. The procedure may last as short a period as the completion of Election Day up to more than two weeks.

106 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-2 September 27, 2005 HAVA requires that states provide individuals casting provisional ballots with free access to a mechanism by which they can determine the disposition of their ballots. The NASS survey revealed that the methods of notification varied among Web sites, toll-free phone lines, and direct contact by local election administrators. Applicability and Coverage HAVA required all states to offer provisional ballots in federal elections beginning in 2004, although some states are exempt because they have no voter registration (North Dakota) or have alternative systems, such as Election Day registration (Idaho, Maine, Minnesota, New Hampshire, Wisconsin, and Wyoming). The states of Maine, Wisconsin, and Wyoming allowed provisional balloting for first-time voters who were unable to provide identification or whose ballots were challenged at the polls. The other four states had no form of provisional balloting and no data was reported for them. New Hampshire and North Dakota allowed voters without identification to sign affidavits swearing to their identity. Minnesota and Idaho did not allow first-time voters without identification to cast ballots. Mississippi and Pennsylvania failed to provide any data on provisional ballot use in their states. The states of California, Indiana, Iowa, Kansas, Missouri, New Mexico, New York, and West Virginia did not provide full statistics for all their jurisdictions. Reasons for rejecting provisional ballots vary. In 18 states, provisional ballots are eligible to be counted if cast outside the individual s home precinct. In 25 states, provisional ballots are disqualified if cast outside the individual s home precinct. States where provisional ballots are eligible to be counted if cast outside the correct precinct: Alaska Delaware New Mexico Utah Arizona Georgia North Carolina Vermont Arkansas Illinois Oregon Washington California Louisiana Pennsylvania Colorado Maryland Rhode Island Historical Context Prior to the adoption of HAVA, some states provided for the casting and counting of provisional ballots. The rules regarding the use of provisional ballots were uneven among states. For example, Ohio provided a method of provisional balloting to persons who moved within the state but did not reregister at their new address by Election Day. California provided a method of provisional balloting to persons who could not establish their eligibility at the polls. Texas provided a method of provisional balloting for persons who were challenged at the polls. State and local jurisdiction statistics are unavailable as to the incidence of these pre-hava forms of provisional balloting. HAVA mandated the use of provisional ballots in federal elections starting January 1, The November 2, 2004, election is the first federal election to be conducted with national usage of provisional ballots. Although provisional balloting has provided a minimum standard for provisional balloting, as described above, the application of how and when provisional ballots will be cast and counted varies among the states. Furthermore, some states permit provisional balloting only in federal elections.

107 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-3 September 27, 2005 Survey Results Table 6 presents data on provisional ballots cast and counted from questions 8 and 9 on the Election Day Survey. In the table, the number of provisional ballots cast is calculated as a percentage of reported total registration, and the number of provisional ballots counted is calculated as a percentage of provisional ballots cast. The column headings in Table 6 are as follows: Column Headings for Table 6. Provisional Ballots Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Registration Number of active and inactive registered voters, number of persons who voted on Election Day in six states, and VAP data for North Dakota and jurisdictions in Wisconsin that do not have voter registration, from col. 4 of Table 2, Registration 5 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data 6 Ballots Cast in Polling Places Number of ballots cast in polling places on Election Day, from col. 9 of Table 4, Turnout Source 7 Cases Number of jurisdictions that responded to survey questions 1 and 3, that provided Election Day registration data, and for which VAP data was substituted for voter registration data 8 Total Provisional Cast Number of provisional ballots cast from survey question 8 9 Cases Number of jurisdictions that responded to question 8 10 Percent Provisional Cast of Registration Number of provisional ballots cast (col. 6) divided by the number of registered voters (col. 4) 11 Cases Number of jurisdictions that responded to survey questions 1 and 8, provided Election Day registration data, or for which VAP data was substituted for voter registration data 12 Cases > 100% Number of jurisdictions where the reported number of provisional ballots cast (col. 6) is greater than the reported number of registered voters (col. 4) 13 Percent Provisional Cast of Polling Places Number of provisional ballots cast (col. 8) divided by the number of ballots cast in polling places on Election Day (col. 6) 14 Cases Number of jurisdictions that responded to survey questions 3 and 8 15 Cases > 100% Number of jurisdictions where the reported number of provisional ballots cast (col. 8) is greater than the number of ballots cast in polling places on Election Day (col. 6) 16 Total Provisional Counted Number of provisional ballots counted from survey question 9 17 Cases Number of jurisdictions that responded to question 9

108 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-4 September 27, 2005 Column Headings for Table 6 (cont.) Col. Heading Description 18 Percent Provisional Counted of Prov Cast Number of provisional ballots counted (col. 11) divided by the number of provisional ballots cast (col. 6) 19 Cases Number of jurisdictions that responded to questions 8 and 9 20 Cases > 100% Number of jurisdictions where the reported number of provisional ballots counted (col. 11) is greater than the reported number of provisional ballots cast (col. 6)

109 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-5 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 6 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the U.S. Election Assistance Commission on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 6 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary Regarding provisional ballots, the Election Day Survey asked for the number of provisional ballots cast, the number counted, and the five most common reasons for rejecting provisional ballots. Overall, at least 1,901,591 individuals sought to cast a provisional ballot in the 2004 election. That amounted to 1.25 percent of all persons registered for the election and 2.56 percent of ballots cast in polling places on Election Day. The states reported that at least 1,225,915 provisional ballots were counted, or percent of those provisional ballots cast. States were also asked to provide the five most common reasons why the provisional ballots were rejected, although the actual numbers of ballots rejected categorized by the reasons for rejection were not requested. The states were not asked to provide this information for their individual jurisdictions, just a statewide summary. The reasons, according to their frequency of mention by states, are as follows: Reasons for Rejecting Provisional Ballots Frequency of Mention Not registered 18 Wrong precinct 14 Improper ID 7 Incomplete ballot form 6 Wrong jurisdiction 5 Already voted 3 Ballot not timely received 3 Ineligible to vote 3 No signature 3 Administrative error 2 Non-matching signature 2

110 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-6 September 27, 2005 Reasons for Rejecting Provisional Ballots (cont.) Frequency of Mention Other 2 Registration purged 2 Deceased 1 Elector challenged 1 First-time voter registering on Election Day 1 Missing ballot 1 Multiple ballots in one envelope 1 Name missing from voter list 1 Nonappearance within 24 hours 1 Nonverifiable signature 1 We calculated three measures of provisional balloting in our analysis: the number of reported provisional ballots cast as a percentage of the voter registration, the number of reported provisional ballots cast as a percentage of ballots cast in polling places, and the report number of provisional ballots that were counted as a percentage of the reported number of provisional ballots cast. Generally we found the same relationships for the number of provisional ballots cast as a percentage of voter registration or as a percentage of ballots cast in polling places. The patterns of provisional balloting revealed by our analysis suggest that administrative rules and procedures are most related to the casting and counting of provisional ballots. Most notably, jurisdictions that permitted jurisdiction-wide acceptance of provisional ballots reported higher rates of provisional ballots being cast, but also reported a much higher incidence of provisional ballots being counted, than other jurisdictions. Those jurisdictions with statewide voter registration databases reported a lower incidence of casting provisional ballots than states without voter registration databases, suggesting that better administration of voter registration rolls might be associated with fewer instances where voters would be required to cast a provisional ballot due to a problem with their voter registration. Over one million provisional ballots were reportedly cast in Section 203 covered jurisdictions, and correspondingly, there was a higher incidence of provisional ballots cast in Section 203 covered jurisdictions than those jurisdictions not covered. The rate of counting the provisional ballots was slightly higher in Section 203 jurisdictions, but could not offset the much higher incidence of casting provisional ballots. On a related note, predominantly Hispanic jurisdictions had the highest rate of casting provisional ballots, followed by predominantly non-hispanic Native American jurisdictions. While the counting of provisional ballots was highest in predominantly Hispanic jurisdictions, predominantly non- Hispanic Native American jurisdictions had a counting rate under 50 percent. Higher incidences of casting provisional ballots can also be found in urban and high population density areas, but these jurisdictions also had higher rates of counting provisional ballots. Rates of counting provisional ballots also tended to increase with the income and education level within a jurisdiction.

111 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-7 September 27, 2005 States Among those jurisdictions reporting, Alaska reported the highest incidence of provisional ballots cast as a percentage of voter registration, at 4.93 percent, followed by California, at 4.08 percent, and Arizona at 3.84 percent. Twenty-four states reported provisional ballots as a percentage of registration at 0.3 percent or lower, with Vermont and Wyoming the lowest at just under 0.03 percent. As a percentage of votes cast at the polling place, Washington was the highest at percent, followed by Alaska, percent; Arizona, 8.99 percent; and California, 8.47 percent. The change in the relative order is a consequence of the varied incidence of other methods of voting, such as absentee and early voting. The states with the lowest incidence of provisional balloting were again Vermont and Wyoming at 0.05 percent. States reported a very wide range of whether the ballots were counted. Maine had the highest rate of provisional ballots counted, and serves as an interesting case, since the state permits first-time voters without required identification to cast a provisional ballot. Maine reports slightly more ballots counted, 486, than cast, 483. This is presumably a consequence of a data-entry error. More generally, nearly all provisional ballots cast in Maine were counted. After Maine, Alaska reported the highest rate of counting provisional ballots, at percent. Thus, even though Alaska had the highest incidence of provisional balloting, those ballots tended to be counted. Oregon followed at 85 percent, and also serves as an interesting case because persons casting a provisional ballot are motivated people who go to their county election administrative offices to cast a ballot if they did not receive one by mail. Washington, Nebraska, and Ohio all reported counting rates near 79 percent. States with low reported rates of counting provisional ballots were Delaware at 6.3 percent, Hawaii at 7.20 percent, and Oklahoma at 7.70 percent. Table 6A shows the states sorted by the two methods of calculations. Regions Jurisdictions in the West reported the highest percentage of provisional ballots cast, 2.94 percent of voter registration or 6.54 percent of votes cast in polling places, but also reported the highest rate of counting those ballots, 74 percent. The Northeast reported the second highest percentage of provisional ballots cast as a percentage of voter registration, 1.34, but reported the lowest rate of counting those ballots, 42.8 percent. As percentage of ballots cast at the polling place, the Northeast reported the lowest incidence of cast provisional ballots at 0.86 percent. The Midwest was next in provisional ballots cast, 0.80 percent of registration or 1.91 of votes cast in polling place, and reported the second highest rate of counting, percent. The South reported the lowest rate of casting provisional ballots, at 0.44 percent of registration and 1.01 percent of ballots cast in polling places, and the third highest rate of counting, at percent. Urban to Rural Urban jurisdictions reported the highest rate of provisional ballots cast, 1.55 percent, followed by suburban jurisdictions at 1.12 percent, small towns at 1.02 percent, and rural jurisdictions, at 0.67 percent. The same pattern was reported when calculated as a percentage of ballots cast in polling places. Suburban jurisdictions reported the highest rate of counting provisional ballots, 73.10

112 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-8 September 27, 2005 percent, followed by rural jurisdictions at percent, urban jurisdictions at percent, and small towns at percent.

113 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-9 September 27, 2005 Table 6a. Provisional Ballot Usage, Sorted

114 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-10 September 27, 2005 Size of Jurisdiction The urban-to-rural trend on rate of provisional ballots cast persists for the population size of the jurisdiction. The reported rate of provisional ballots cast increases with population size, from 0.10 percent for voter registration in jurisdictions under 1,000 voting age population (VAP), to 2.51 percent in jurisdictions over one million VAP. For percentage of ballots cast in polling places, the percentages ranged from 0.08 percent for the smallest jurisdiction to 6.08 percent for the largest. The reported rate of counting provisional ballots generally increased with population size of the jurisdiction. Jurisdictions with a population between 1,000 and 3,500 VAP reported the lowest rate of counting provisional ballots, at percent, while the largest jurisdictions reported percent. However, the trend did not hold for the smallest jurisdictions below 1,000 VAP, which reported percent. Race and Ethnicity The highest reported incidence of casting provisional ballots among voter registration was in predominantly Hispanic jurisdictions, 2.81 percent, followed by predominantly non-hispanic Native American jurisdictions, 1.89 percent; predominantly non-hispanic Black areas, 1.28 percent; and predominantly non-hispanic White communities, 1.12 percent. The order was the same when calculated as a percentage of ballots cast in polling places, ranging between 6.25 and 2.25 percent. The highest reported rate of counting provisional ballots was also among predominantly Hispanic jurisdictions, percent, followed by predominantly non-hispanic White areas, percent; predominantly non-hispanic Black communities, percent; and predominantly non-hispanic Native American jurisdictions, percent. Median Income Reported rates of casting provisional ballots as a percentage of voter registration generally rise with the income level of the jurisdiction, from a 0.22 percent rate for jurisdictions with a median income less than $25,000 to a 1.52 percent rate for median income between $40,000 and $45,000. The rate drops off for the highest income jurisdictions, to 1.29 percent for those with a median income above $50,000. The same pattern holds when calculated as a percentage of ballots cast in polling places, ranging from 0.63 percent for the lowest category to 3.22 percent for jurisdictions with $40,000 $45,000 median income, before dropping slightly to 2.49 percent for the highest category. Generally, higher income jurisdictions counted provisional ballots at nearly twice the rate of lower income communities. The reported rates for counting ballots follows a similar pattern, from a low of percent counted in the lowest income category, to a high in the $45,000 and $50,000 range of percent, and then dropping off slightly for the highest income category to percent. However, jurisdictions in the $35,000 and $40,000 range break the pattern, dipping to a percent counted rate. High School Education The greatest variation in reported rates of provisional ballots cast occurs for the two lowest education categories. For the lowest, the rate of casting ballots is 0.23 percent; the rate jumps to 2.37 percent in the next highest category, and then declines to a little more than 1.00 percent for the remainder. The pattern is the same when calculated as a percentage of ballots cast in polling places,

115 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-11 September 27, percent for the lowest category, 5.41 for the next highest, and about 2.00 percent for the remainder. The reported counting rate of provisional ballots generally shows a positive relationship between ballots counted and education levels, rising from a low of percent for the lowest education category and rising to percent for the highest. The deviation from the increasing pattern occurs at the medium range of percent high school education, with a counting rate dipping to percent. Section 203 Language Minority Requirements There is a large difference in the reported rate of provisional ballots cast among Section 203 covered jurisdictions. Those covered reported a rate based on voter registration much higher, 2.04 percent, than those that are not covered, 0.82 percent. When calculated as a percentage of ballots cast in polling places, Section 203 covered jurisdictions reported 5.09 versus 1.38 percent for other jurisdictions. Section 203 covered jurisdictions reported a slightly higher rate of counting provisional ballots, 68.4 percent, than those not covered, 59.8 percent. Section 5 Preclearance of Voting Procedures Section 5 covered jurisdictions reported a slightly lower rate of casting provisional ballots when calculated as a percentage of voter registration, 1.03 versus 1.25 percent, and a slightly higher rate when calculated for ballots cast in polling places, 2.49 versus 2.42 percent. Section 5 covered jurisdictions reported a slightly higher rate of counting provisional ballots than jurisdictions not covered by Section 5, versus percent. Type of Voting Equipment Among those jurisdictions reporting voting equipment, those with lever machines reported the highest rate of casting provisional ballots when calculated as a percentage of voter registration, at 1.61 percent, but the second lowest when calculated as a percentage of ballots cast in polling places, 0.68 percent. Paper jurisdictions reported the lowest rate by either measure, 0.30 percent for voter registration and 0.39 percent for ballots cast in polling places. Most jurisdictions using other types of voting equipment have similar rates of casting provisional ballots, around 1 percent for voter registration or 2 to 3 percent for ballots cast in polling places. Lever machine jurisdictions reported the lowest rate of counting those ballots, percent, followed by paper jurisdictions, which reported a counting rate of 58 percent. Other jurisdictions ranged between 60 and 70 percent counting rates. Changed Voting Equipment since 2000 Those jurisdictions that changed voting equipment reported a higher rate of provisional ballots cast 1.54 percent for voter registration and 3.42 percent for ballots cast in polling places than those jurisdictions that did not, which measured 1.05 and 1.97 percent, respectively. Those jurisdictions that changed voting equipment also reported a higher rate of provisional ballots counted, percent, than those jurisdictions that did not, percent.

116 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-12 September 27, 2005 Statewide Voter Registration Database Statewide voter registration databases lead to almost half the number of provisional ballots being cast. Those jurisdictions with statewide voter registration databases reported a lower rate of casting provisional ballots, 0.59 percent for voter registration and 1.21 percent for ballots cast in polling places, than other jurisdictions, 1.37 and 2.86 percent, respectively. Both types of jurisdictions reported similar levels of counting provisional ballots, slightly above 64 percent. Election Day Registration Those jurisdictions with Election Day registration might reasonably be assumed to not need provisional ballots because voters can register at the polls. However, for three of the seven Election Day registration states Maine, Wisconsin, and Wyoming provisional balloting was provided for first-time voters who were unable to provide identification or voters whose ballots were challenged at the polls. As the numbers show, this was a rare event in these three states. In those states with Election Day registration the reported incidence of provisional ballots cast was 0.03 percent or registration or 0.04 percent of ballots cast in polling places, and 78 percent of these ballots were counted. For states without Election Day registration, the reported incidence of provisional ballots cast was 1.22 percent of registration or 2.50 percent of ballots cast in polling places, and 64.3 percent were counted. Provisional Ballot Acceptance Jurisdictions with jurisdictionwide provisional ballot acceptance reported higher rates of provisional ballots cast, 2.09 percent of registration or 4.67 percent of ballots cast in polling places, than those with in-precinct-only acceptance, 0.72 and 1.18 percent, respectively. Predictably, those jurisdictions with more permissive jurisdictionwide acceptance reported higher rates of counting provisional ballots, percent, than other jurisdictions, percent. No Excuse Absentee Balloting Jurisdictions with no excuse absentee balloting reported more than twice the rate of casting provisional ballots, 1.94 percent of registration or 4.20 of ballots cast in polling places, than those jurisdictions that did not, 0.74 and 1.14 percent, respectively. Jurisdictions with no excuse absentee balloting reported a higher rate of counting provisional ballots, 71.7 percent, than those jurisdictions that did not, 52.5 percent. Early Voting Jurisdictions with early voting reported a higher incidence of provisional ballots cast, 1.52 percent of registration and percent of ballots cast in polling places, than those jurisdictions that did not, 0.93 and 1.45 percent, respectively. Jurisdictions with early voting reported a higher rate of provisional ballots counted, percent, compared to other jurisdictions, percent. Battleground States Jurisdictions in battleground states reported a slightly lower incidence of casting provisional ballots, 1.04 percent pf registration and 2.39 of ballots cast in polling places, than those jurisdictions that were not battleground states, which measured 1.27 and 2.46 percent, respectively. Jurisdictions in battleground states reported a higher rate of counting provisional ballots, percent, than those jurisdictions that were not in battleground states, at percent.

117 2004 Election Day Survey Report, Part 2 Survey Results Provisional Ballots, Page 6-13 September 27, 2005 Presidential Margin of Victory No clear pattern emerges for provisional balloting and presidential margin of victory. The reported incidence of casting provisional ballots ranged from 0.78 to 1.32 percent of registration and 1.68 to 2.86 percent of ballots cast in polling places. The reported rate of counting provisional ballots ranged from to percent. Of note, the range where either presidential candidate won by between 5.00 to 7.50 percent reported both the highest incidence of provisional ballot casting and rate of counting. Red versus Blue Jurisdictions Jurisdictions in which Bush won a plurality of the vote reported the lowest incidence of casting provisional ballots, 0.37 percent of registration and 0.51 percent of ballots cast in polling places, while those that were won by Kerry by more than 55 percent reported the highest incidence of casting provisional ballots, 1.65 and 3.28 percent, respectively. For the other categories, the incidence of casting provisional ballots was slightly less than 1 percent for registration and around 2 percent for ballots cast in polling places. Jurisdictions in which Bush won a plurality of the vote reported the lowest rate of counting provisional ballots, 54.5 percent, while those that were won by Bush with between 50 and 55 percent reported the highest rate of counting provisional ballots, 73.2 percent. Those that were won by Kerry by more than 55 percent reported the second highest rate of counting ballots, at 71.0 percent. The remainder of jurisdictions varied between 59.9 percent and 68.2 percent. REFERENCES National Association of Secretaries of State Summaries and Highlights: NASS Survey of the Election Community Regarding Provisional Ballots. Washington, DC.

118 EAC Election Day Survey Provisional Ballots 2004 General Election Provisional Provisional Ballots Cast Cases = Number of Jurisdictions Reporting Subject Matter Provisional Ballots Counted Updated: 09/19/ :04:16 Ballots Percent Percent Percent Election Cast Total Provisional Provisional Total Provisional Administration Total In Polling Provisional Cast of Cases Cast of Cases Provisional Counted of Cases Code Name Jurisdictions Registration Cases Place Cases Cast Cases Registration Cases >100% Polling Place Cases >100% Counted Cases Prov Cast Cases >100% 01 Alabama 67 2,597, , , Alaska 1 472, , , , Arizona 15 2,642, ,129, , , Arkansas 75 1,699, , , , California 58 16,646, ,920, , , Colorado 64 3,101, , , , Connecticut 169 1,831, ,452, , Delaware 3 553, , District of Columbia 1 383, , , , Florida 67 10,300, ,865, , , Georgia 159 4,248, ,642, , , Hawaii 5 647, , Idaho , , Illinois 110 7,195, , , Indiana 92 4,296, ,251, , Iowa 99 2,226, ,073, , , Kansas 105 1,695, , , , Kentucky 120 2,794, , Louisiana 64 2,932, ,801, , , Maine 517 1,026, , Maryland 24 3,105, ,222, , , Massachusetts 351 4,098, ,821, , , Michigan 83 7,164, ,250, , , Minnesota 87 2,977, ,611, Mississippi 82 1,469, Missouri 116 4,194, , , Montana , , Nebraska 93 1,160, , , , Nevada 17 1,073, , , , New Hampshire , , New Jersey 21 5,011, ,409, , , New Mexico , , , , New York 58 11,837, , , North Carolina 100 5,526, ,413, , , North Dakota , , Ohio 88 7,965, ,995, , , Oklahoma 77 2,143, ,324, , Oregon 36 2,141, ,585, , , Pennsylvania 67 8,366, Rhode Island , , , South Carolina 46 2,318, , , South Dakota , Tennessee 95 3,748, ,297, , , Texas ,098, ,641, , , Utah 29 1,278, , , , Vermont , , Virginia 134 4,515, ,001, , Washington 39 3,508, , , , West Virginia 55 1,168, , , , Wisconsin 1,910 4,179,774 1, Wyoming , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 2,440, ,947, , , Virgin Islands 1 50, , Total 6, ,265,030 6,512 67,603,992 3,850 1,901,591 4, , , ,225,915 4, , Maximum 1,910 16,646,555 1,894 7,920, , , Average 119 3,344, ,572, , , Minimum 1 50, , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

119 EAC Election Day Survey Provisional Ballots 2004 General Election Provisional Provisional Ballots Cast Cases = Number of Jurisdictions Reporting Subject Matter Provisional Ballots Counted Updated: 09/19/ :04:16 Ballots Percent Percent Percent Election Cast Total Provisional Provisional Total Provisional Administration Total In Polling Provisional Cast of Cases Cast of Cases Provisional Counted of Cases Code Name Jurisdictions Registration Cases Place Cases Cast Cases Registration Cases >100% Polling Place Cases >100% Counted Cases Prov Cast Cases >100% Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,484, ,286, , , Punch card ,767, ,875, , , Lever ,662, ,700, , , Paper 1,734 3,085,167 1,733 1,044,700 1,011 6, , Optical scan 2,541 69,198,628 2,523 28,352,237 1, ,694 1, , , ,380 1, ,433 2 Electronic ,068, ,384, , , Multiple Systems ,997, ,959, , , Changed Voting Equipment Since 2000 General Election Yes 1,753 51,149,755 1,746 21,652, , , No 4, ,115,275 4,766 45,951,575 2,976 1,173,874 3, , , ,386 3, ,421 2 State Wide Voter Registration System in Place Yes 1,335 37,384,852 1,321 19,051,011 1, ,421 1, , , ,619 1, No 5, ,880,178 5,191 48,552,981 2,761 1,698,170 2, , , ,094,296 2, ,082 3 Election Day Registration Yes 2,823 10,323,368 2,806 4,701, No 3, ,941,662 3,706 62,902,429 2,938 1,900,639 3, , , ,225,286 3, ,830 7 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,077,741 1,143 23,631, ,178,607 1, , ,145 1, In Precinct Only 4, ,336,604 4,316 37,233,762 1, ,807 2, , , ,562 2, ,962 3 None 1,056 8,850,685 1,053 6,739,037 1,053 22, , No Excuse Absentee Balloting Yes 3,781 64,333,790 3,750 27,451,170 1,746 1,172,134 1, , , ,965 1, ,067 3 No 2, ,931,240 2,762 40,152,822 2, ,457 2, , , ,950 2, ,910 5 Early Voting Allowed Yes 1,701 73,710,075 1,686 32,353,422 1,657 1,106,561 1, , , ,108 1, ,260 6 No 4, ,554,955 4,826 35,250,570 2, ,030 2, , , ,807 2, ,717 2 Covered By Section 203, Language Minority Requirements Yes ,756, ,294, ,002, , No 6, ,508,534 6,059 49,309,139 3, ,774 3, , , ,518 3, ,584 7 Covered By Section 5 of Voting Rights Act Yes ,868, ,774, , , No 5, ,396,175 5,648 51,829,587 3,169 1,496,329 3, , , ,510 3, ,290 4 Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

120 EAC Election Day Survey Provisional Ballots 2004 General Election Provisional Provisional Ballots Cast Cases = Number of Jurisdictions Reporting Subject Matter Provisional Ballots Counted Updated: 09/19/ :04:16 Ballots Percent Percent Percent Election Cast Total Provisional Provisional Total Provisional Administration Total In Polling Provisional Cast of Cases Cast of Cases Provisional Counted of Cases Code Name Jurisdictions Registration Cases Place Cases Cast Cases Registration Cases >100% Polling Place Cases >100% Counted Cases Prov Cast Cases >100% Demographics Region Northeast 1,710 34,273,670 1,709 9,736,138 1, ,060 1, , , ,813 1, South 1,423 62,606,676 1,407 25,158,163 1, ,037 1, , , ,361 1, ,185 5 Midwest 2,902 44,048,138 2,879 16,057, , , West ,845, ,674, , , Territories 113 2,490, ,977, , , Urban to Rural Urban ,441, ,932, , , Suburban ,552, ,338, , , Small Towns 1,710 44,193,768 1,690 15,783,352 1, ,198 1, , , ,128 1, Rural 3,307 19,586,556 3,277 7,571,710 1, ,162 1, , , ,485 1, ,262 6 Not Available - Territories 113 2,490, ,977, , , Size of Jurisdiction (VAP) < 1,000 1, ,006 1, , >=1,000 to <3,500 1,165 2,182,148 1, , , , >=3,500 to <10,000 1,043 5,966,645 1,037 2,618, , , >=10,000 to <50,000 1,704 31,472,681 1,681 12,888,120 1, ,002 1, , ,166 69,882 1, ,321 4 >=50,000 to <250, ,992, ,301, , , >=250,000 to <1,000, ,396, ,669, , , >=1,000, ,867, ,147, , , Not Available 144 2,492, ,978, , , Race and Ethnicity Predominently NH White 6, ,662,585 6,234 60,592,039 3,627 1,622,859 3, , , ,017,790 3, ,761 7 Predominently NH Black 85 3,098, ,460, , , Predominently NH Native American , , , , Predominently Hispanic 50 7,749, ,480, , , Not Available 145 2,523, ,987, , , Median Income < $25, ,504, , , , >=$25,000 to <$30, ,917, ,268, , , >=$30,000 to <$35,000 1,372 22,970,583 1,366 7,393, , , >=$35,000 to <$40,000 1,215 40,443,694 1,213 11,981, , , >=$40,000 to <$45, ,780, ,288, , , >=$45,000 to <$50, ,218, ,091, , , >=$50,000 1,180 40,936,586 1,178 17,994, , , Not Available 151 2,492, ,978, , , High School Education < 60% 126 1,817, , , , >=60% to <70% ,944, ,195, , , >=70% to <80% 1,646 49,285,773 1,631 15,020, ,390 1, , ,695 1, >=80% to <90% 3,111 93,198,279 3,105 36,448,287 1, ,040 1, , , ,535 1, ,197 3 >=90% ,495, ,496, , , Not Available 151 2,523, ,987, , , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

121 EAC Election Day Survey Provisional Ballots 2004 General Election Provisional Provisional Ballots Cast Cases = Number of Jurisdictions Reporting Subject Matter Provisional Ballots Counted Updated: 09/19/ :04:16 Ballots Percent Percent Percent Election Cast Total Provisional Provisional Total Provisional Administration Total In Polling Provisional Cast of Cases Cast of Cases Provisional Counted of Cases Code Name Jurisdictions Registration Cases Place Cases Cast Cases Registration Cases >100% Polling Place Cases >100% Counted Cases Prov Cast Cases >100% Political Battleground States in 2004 Presidential Election Yes 3,093 64,166,639 3,062 23,916, , , No 3, ,098,391 3,450 43,687,838 2,890 1,396,522 3, , , ,061 3, ,216 6 Margin of Victory in 2004 Presidential Election < 2.5% ,923, ,750, , , >=2.5% to < 5.0% ,133, ,999, , , >=5.0% to < 7.5% ,830, ,578, , , >=7.5% to < 10.0 % 429 8,833, ,452, , , >=10.0 % 4, ,044,988 4,463 46,839,469 2,664 1,444,826 2, , , ,002 2, ,196 8 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,178,580 3,094 26,081,362 1, ,309 2, , , ,479 2, ,672 7 Bush 50% to 55% ,682, ,749, , , Bush < 50% 136 2,041, , , , Kerry < 50% 150 4,850, ,743, , , Kerry 50% to 55% ,160, ,518, , , Kerry > 55% 1,161 49,846,628 1,154 18,869, , , Tied 25 14, , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

122 Chapter 7 Drop-Off Table 7 provides data from the Election Day Survey on drop-off rates. Drop-off is the difference between voter turnout (total ballots cast) and the total number of votes cast for all candidates in a particular contest. This raw number difference is usually expressed as a percentage of the total votes cast in the election. For example, if one hundred people turned out to vote, and ninety of them cast a ballot for President, there would be a 10 percent drop-off for President. The U.S. Election Assistance Commission (EAC) asked for the total number of votes cast for the three federal offices that were on the ballot in the 2004 general election (U.S. president, U.S. Senate, and U.S. House of Representative), and therefore drop-off is calculated for only those contests. On the other hand, drop-off can be calculated for any office on the ballot, all the way down to local contests and referendums. Generally, the farther down the ballot, the higher the rate of drop-off as voter fatigue or unfamiliarity with the candidates or issues increases. Drop-off represents a combination of overvotes and undervotes, which we analyze in chapter 8, and is also sometimes referred to as the residual vote. Drop-off rates for each office in this study were calculated from survey questions on ballots cast and votes for all candidates in each federal contest. Applicability and Coverage Citizens of the territories of the United States cannot cast votes for president and Senate, but do have nonvoting representation in the U.S. House. Presidential vote totals were not received for the states of Pennsylvania and South Carolina, and U.S. House results were not received from Illinois, Ohio, Pennsylvania, South Carolina, and Tennessee. In calculating drop-off from the numbers reported to the EAC through the Election Day Survey, Election Data Services identified 903 jurisdictions that reported zero drop-off for president. For some of these jurisdictions, particularly smaller jurisdictions, this may be a correct number. On the other hand, for jurisdictions with larger populations, this zero drop-off is likely a consequence of jurisdictions historically reporting the total votes for highest office as the total turnout. For 2004, the entire state of Arkansas, a significant number of jurisdictions in Mississippi and Vermont, along with numerous individual jurisdictions in 21 other states followed this practice. It is also possible that some jurisdictions misinterpreted this survey item. In addition, Election Data Services also found 176 jurisdictions that reported a negative drop-off for president, which cannot be logically correct since it implies that more people voted for president than cast a ballot. Research into a number of these negative drop-off jurisdictions found data-entry errors in the answers submitted by jurisdictions to the Election Day Survey. Unfortunately, we did not have the resources to validate every number. For contests for U.S. Senate, 412 jurisdictions reported zero drop-off and 138 reported negative drop-off. For U.S. House, 372 jurisdictions reported zero drop-off and 72 reported negative drop-off. The error is correlated across offices on the ballot. Jurisdictions reporting zero drop-off for president,

123 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-2 September 27, 2005 Senate, and U.S. House totaled 302. Forty-five jurisdictions reported negative drop-off for president, Senate, and U.S. House. Care should also be used in interpreting the drop-off data calculated for the U.S. House. Some jurisdictions have more than one House district, and summing drop-off across districts appears to have been a high administrative hurdle, as many large population jurisdictions reported implausibly high drop-off for U.S. House. In cases where a candidate ran unopposed, some jurisdictions save the administrative costs of holding the election and simply declare the unopposed candidate the winner. Drop-off in these jurisdictions will be much higher, and will be further confounded if a jurisdiction had two districts where one race was contested and one was not. However, it may not always be the case that the presidential election will have the smallest drop-off. We note that jurisdictions tended to report less drop-off for Senate or U.S. House than president, particularly in hotly contested lower ballot elections. For example, all but one of South Dakota s counties reported less drop-off for Senate than for president. In all, 272 jurisdictions reported less drop-off in the Senate race than in the presidential race and 188 jurisdictions reported less drop-off in the U.S. House race than in the presidential race. Six hundred fifty-five jurisdictions reported less drop-off in the U.S. House race than in the Senate race. That Senate or U.S. House turnout can be higher than presidential turnout explains some of the negative values for Senate drop-off among jurisdictions that reported presidential drop-off as total ballots cast. Thirty-seven jurisdictions with zero presidential drop-off reported negative Senate dropoff and 10 jurisdictions with zero reported presidential drop-off reported negative U.S. House dropoff. Historical Context Not all persons register a vote for a particular office on the ballot, even if it is the first contest listed. Some abstention is intentional, where a voter may feel they do not know enough about the candidates or issues on the ballot in order to cast a vote in the particular contest. Some abstention may be due to voter error by failing to mark a ballot so that a vote can be recorded, or by casting a vote more times than allowed. The state of Nevada has attempted to cut down the level of drop-off by providing a separate ballot line for none of these candidates in the presidential, U.S. Senate, judgeship, and other statewide contests. This ballot line is treated as if the contest had another candidate. But, despite this effort, there is still drop-off in these contests. For example, in the 2004 presidential contest, 3,688 or 0.44 percent of voters cast a vote for none of these candidates, but another 1,976 voters (or 0.24 percent) failed to register any vote for the presidential office. In the past, Election Data Services has provided election statistics to the Congressional Research Service (Crocker 1996). Among these statistics are two numbers that are related to drop-off: the total number of ballots cast and the vote for highest office, which tend to be the vote for president in a presidential election year or the vote for governor, U.S. Senate, or the summation of all U.S. House races in nonpresidential years. Election Data Services calculates the vote for highest office for each jurisdiction in the state, which leads to variation in which office is used across the state. The

124 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-3 September 27, 2005 coverage of the total ballots cast varies. Seventeen states reported the total number of ballots cast in 1948, half in 1970, and 38 in Figure 7a plots vote for highest office drop-off in federal elections, post-world War II, from (The data are drawn from Table 3a in this report.) There has been little in the way of a trend in percent drop-off over this time period, with drop-off generally remaining between 1.5 percent and 3.5 percent. It is noteworthy that residual vote is lower in presidential elections than in midterm elections. One factor that is related to the narrowing of the residual vote is the competition for the race at the top of the ballot. During a presidential election, voters are drawn to the high profile election and fewer abstain from the race at the top of the ballot. In midterm elections, sometimes one of the two parties will field a weak candidate or even no candidate, and thus some voters may abstain from this race, but choose to participate in another contest on the same ballot. In the 2004 election, reported drop-off was 0.99 percent, the lowest level in post-world War II elections. As we shall see, drop-off is lowest in jurisdictions using electronic voting, so this may be a consequence of the increasing adoption of that voting technology. The 2004 election was also a close election, and voters were primed to believe that their vote counted more than in other elections and, as a consequence of the aftermath of the 2000 Florida recount, were told to closely pay attention to their vote in order to make sure that it was properly recorded. Greater attention to the casting of ballots by voters and the new technology may have contributed to the low drop-off rate in the 2004 election.

125 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-4 September 27, % Figure 7.1 Drop-Off Rate, % 3.5% 3.0% Drop-Off Rate 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% Year

126 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-5 September 27, 2005 Survey Results Table 7 presents drop-off data for federal elections covered by question 12 on the Election Day Survey. In the table, the drop-off rates for presidential, senatorial, and congressional elections are calculated as percentages of ballots counted. The column headings in Table 7 are as follows: Table 7 Column Headings. Drop-Off Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Ballots Counted Number of ballots counted from survey question 2 5 Cases Number of jurisdictions that responded to question 2 6 Total President Number of votes for President from survey question 12 7 Cases Number of jurisdictions that responded to question 12 8 Number Not Voting for President 9 Percent Drop- Off for President Number of ballots counted (col. 4) minus the number of votes for President (col. 6) Number of votes for President (col. 6) divided by the number of ballots counted (col. 4). 10 Cases Number of jurisdictions that responded to questions 2 and Total Pres. > Total Ballots Exceptions Number of jurisdictions that reported more votes for President (col. 6) than the total number of ballots counted (col. 4) 12 Total Senate Number of votes for U.S. Senator from survey question Cases Number of jurisdictions that responded to question Number Not Voting for Senate 15 Percent Drop- Off for Senate Number of ballots counted (col. 4) minus the number of votes for U.S. Senator (col. 12) Number of votes for U.S. Senator (col. 12) divided by the number of ballots counted (col. 4). 16 Cases Number of jurisdictions that responded to questions 2 and Total Sen. > Total Ballots Exceptions Number of jurisdictions that reported more votes for U.S. Senator (col. 12) than the total number of ballots counted (col. 4) 18 Total Cong. Dist. Number of votes for U.S. Representative from survey question Cases Number of jurisdictions that responded to question Number Not Voting for Cong. Dist. 21 Percent Drop Off for Cong. Dist. Number of ballots counted (col. 4) minus the number of votes for U.S. Representative (col. 18) Number of votes for U.S. Representative (col. 18) divided by the number of ballots counted (col. 4). 22 Cases Number of jurisdictions that responded to question 2 and Total Cong. > Total Ballots Exceptions Number of jurisdictions that reported more votes for U.S. Representative (col. 18) than the total number of ballots counted (col. 4)

127 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-6 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 7 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 7 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary Drop-off is the difference between voter turnout (ballots cast) and the total number of votes for all candidates in a contest. Drop-off is reported for three federal offices: president, Senate, and U.S. House. Lacking resources to validate all these data, we exclude jurisdictions reporting negative dropoff for the state-level responses to the Election Day Survey as presented in Table 7. For the subtotaling tabulations appearing below the state statistics in Table 7, we remove jurisdictions reporting zero drop-off. We recognize that by doing so, we are likely inflating the amount of drop-off reported in our tabulations. After examination of the responses to the Election Day Survey, we believe that a significant number of jurisdictions reported presidential vote as total ballots cast, and thus reported zero drop-off for president and that including those jurisdictions leads to shifts in the drop-off analysis. Although Arkansas and major parts of Mississippi and Vermont were reporting presidential vote as the total ballots cast for all jurisdictions, there are many jurisdictions in other states that appeared to do the same. The analysis is thus biased either by keeping zero values for drop-off reported values or by excluding all zero values. We choose to exclude the zero values because we believe it to be in less error. There is a pattern evident across offices found in many academic studies of drop-off, and an associated measure known as roll-off, or the the tendency of the electorate to vote for prestige offices but not for the lower offices on the same ballot (Burnham 1965: 9). Drop-off was least for the presidential election, reported 1.02 percent; higher for the lower profile Senate races, reported 6.86 percent; and highest for the House races, reported percent. The primary draw for voters is information and excitement about the election, and academic studies consistently find that voters have more information and follow the presidential election more closely than Senate and U.S. House races.

128 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-7 September 27, 2005 There is some evidence that competition reduced drop-off. Those jurisdictions with a closer presidential margin of victory reported lower rates of presidential drop-off. However, Senate and U.S. House reported drop-off was not related to the presidential margin of victory, but was associated with slightly lower drop-off for these offices among battleground states. Among the demographic tabulations, reported presidential, Senate, and U.S. House drop-off is related to education and income levels, with lower levels of education and income related to higher rates of drop-off. Presidential drop-off is reportedly high in predominantly non-hispanic Native American jurisdictions, and is more than twice the drop-off in predominantly Hispanic jurisdictions and nearly eight times greater than predominantly non-hispanic Black jurisdictions. Interestingly, Senate and U.S. House drop-off is least in non-hispanic Native American jurisdictions, likely a consequence of the hotly contested Senate and U.S. House races in South Dakota. Section 203 and Section 5 jurisdictions reported large drop-off in U.S. House elections, perhaps because these jurisdictions are located within heavily Democratic districts that rarely draw a strong Republican challenger. Among types of voting equipment, paper and punch card jurisdictions report about 50 percent more drop-off than optical scan jurisdictions and twice the presidential drop-off of all other jurisdictions. States In the presidential election, New Mexico reported the largest presidential drop-off, 2.61 percent, but caution should be used in interpreting this number, since only two-thirds of the counties in the state reported information. Excluding Arkansas, Vermont reported the lowest drop-off rate of.02 percent, but here too incorrect data provided by many towns in the state makes the statewide number suspect. For the Senate elections, the competitiveness of the election is related to drop-off. For example, in Idaho, Republican Crapo ran unopposed, except for a small number of write-in votes for Democrat McClure. Although Crapo won a landslide victory, many Idaho voters chose to abstain, and the state reported drop-off in the Senate election of percent. In contrast, South Dakota, with a closely contested Senate campaign, reported a smaller drop-off for Senate than for president. Between-state comparisons of U.S. House drop-off are less reliable due to the uneven reporting across the states. Regions Across regions, the Midwest and West reported higher drop-off, 1.34 and 1.22, respectively, for president than the Northeast and South,.085 and.096, respectively. The pattern is not the same for the Senate and U.S. House races, primarily because these strongly contested races are different than the presidential battleground states. Few contested Senate elections were held in the Northeast, which reported the highest Senate drop-off, 9.73 percent. All other jurisdictions reported less than half the Senate drop-off of the Northeast, between 3.32 and 4.50 percent. Urban to Rural Rural jurisdictions reported the highest percentage of presidential drop-off, 1.60 percent, while other jurisdictions varied between 0.95 and 1.19 percent. For Senate, urban jurisdictions reported the highest percent of drop-off, 5.32 percent, while the remainder varied between 4.27 and 4.34 percent.

129 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-8 September 27, 2005 Size of Jurisdiction Presidential drop-off tended to decrease with increasing jurisdiction population, with jurisdictions in the second smallest category, 1,000 to 3,500 voting age population (VAP), reporting the highest presidential drop-off, 2.82 percent, while jurisdictions in the 250,000-to-1 million range reported the lowest drop-off, 0.78 percent. There is no clear pattern related to the population size of a jurisdiction for reported drop-off for Senate, varying between 3.38 and 8.56 percent, and U.S. House races, varying between 7.66 and percent. Race and Ethnicity Predominantly non-hispanic Native American jurisdictions reported the highest presidential dropoff, 4.18 percent, over five times that of predominantly non-hispanic Black jurisdictions, at 0.82 percent. Hispanic jurisdictions reported the second highest drop-off, 1.17 percent, followed by predominantly non-hispanic White, 1.07 percent. In contrast to president, non-hispanic Native American jurisdictions reported the lowest level of drop-off for the Senate, 3.30, due perhaps to the high profile election in South Dakota, and the lowest for U.S. House, 4.65 percent, again due in part to a high profile U.S. House race in South Dakota. Predominantly Hispanic jurisdictions reported the highest percentage of drop-off in Senate elections, 5.48 percent. African American jurisdictions had the highest rate of drop-off in U.S. House races, percent, likely due to the noncompetitive nature of congressional races in heavily African American districts. Median Income There is a strong pattern to drop-off in jurisdictions according to their income levels. Those with the lowest median income completion have more than three times the level of presidential drop-off than those jurisdictions with the highest median income, 2.41 versus 0.78 percent. The same pattern of decreasing reported drop-off with rising income generally holds for Senate elections, 7.33 for the lowest income areas versus 4.32 percent for the highest income areas, but no relationship was evident in U.S. House elections. High School Education There is a strong pattern to drop-off in jurisdictions according to their education levels. Those with the lowest rates of high school completion have nearly four times the amount of presidential drop-off than those jurisdictions with the highest rate of high school completion, 2.05 versus 0.69 percent. The same pattern of reported decreasing drop-off with rising education generally holds for Senate, 7.53 versus 3.75 percent, but no clear relationship was evident in U.S. House elections. Section 203 Language Minority Requirements Section 203 covered jurisdictions reported slightly higher presidential drop-off than other jurisdictions, 1.17 versus 1.07 percent. Section 203 covered jurisdictions reported slightly lower Senate drop-off compared with other jurisdictions. Section 203 covered jurisdictions reported higher U.S. House drop-off than other jurisdictions, 4.60 versus 8.01 percent, and reported higher drop-off in U.S. House elections, versus percent. Section 5 Preclearance of Voting Procedures Section 5 jurisdictions reported slightly higher presidential drop-off than other jurisdictions, 1.19 versus 1.07 percent. Section 5 jurisdictions reported slightly lower Senate drop-off than other

130 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-9 September 27, 2005 jurisdictions, 4.30 versus 7.55 percent. Section 5 jurisdictions reported higher U.S. House drop-off than other jurisdictions, versus percent. Type of Voting Equipment Among known types of voting equipment, punch card and paper equipment have higher presidential drop-off than other types, 1.60 percent and 1.54 percent, respectively. Optical scan equipment has the next highest drop-off at around 1.12 percent, followed by electronic, lever, and multiple systems jurisdictions, all reporting slightly higher than 0.8 percent. Drop-off rates for Senate and U.S. House are uniformly higher among all types of voting equipment. Jurisdictions with lever machines reported the highest drop-off for both Senate and U.S. House, at 9.81 percent and percent, respectively. Other jurisdictions reported similar drop-off rates for Senate, ranging from 3.5 percent to 4.3 percent. (Multiple systems reported drop-off of 2.31 percent.) There was more variation in U.S. House drop-off, with jurisdictions using punch cards reporting the lowest drop-off, 4.38 percent. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment reported slightly lower presidential drop-off than those that did not, 0.97 versus 1.16 percent. Jurisdictions that changed voting equipment reported lower Senate drop-off than those that did not, 3.77 versus 8.67 percent, and higher U.S. House drop-off than other jurisdictions, versus percent. Statewide Voter Registration Database Jurisdictions with a statewide voter registration database reported lower presidential drop-off than those that did not, 0.82 versus 1.18 percent. Jurisdictions with a statewide voter registration database reported lower Senate drop-off than those that did not, 4.70 versus 7.45 percent. Jurisdictions with a statewide voter registration database reported higher U.S. House drop-off than those that do not have one, versus percent. Election Day Registration Jurisdictions with Election Day registration reported higher drop-off for presidential, 1.34 versus 1.08 percent, and senatorial contests, 7.74 versus 6.97 percent, than those that do not have Election Day registration. Jurisdictions with Election Day registration reported lower U.S. House drop-off than those that do not, 5.75 versus percent. Provisional Ballot Acceptance There was slightly higher reported presidential drop-off among jurisdictions that accept provisional ballots jurisdiction-wide than those that accept provisional ballots only cast within precinct, 1.14 versus 1.07 percent. There was slightly higher reported Senate drop-off among jurisdictions that accept provisional ballots jurisdiction-wide than those that accept provisional ballots cast only within precinct, 3.87 versus 5.29 percent. There was slightly lower reported U.S. House drop-off among jurisdictions that accept provisional ballots jurisdiction-wide than those that accept provisional ballots only cast within precinct, versus percent.

131 2004 Election Day Survey Report, Part 2 Survey Results Drop-Off, Page 7-10 September 27, 2005 No Excuse Absentee Balloting Jurisdictions with no excuse absentee balloting reported higher presidential drop-off than those that do not, 1.18 versus 1.04 percent. Jurisdictions with no excuse absentee balloting reported slightly lower Senate drop-off than those that do not, 4.21 versus percent. Jurisdictions with no excuse absentee balloting reported lower U.S. House drop-off than those that do not, versus percent. Early Voting Jurisdictions with early voting reported slightly higher levels of presidential drop-off than other jurisdictions. Jurisdictions with early voting reported lower levels of Senate drop-off than other jurisdictions. Jurisdictions with early voting reported slightly higher levels of U.S. House drop-off than other jurisdictions. Battleground States Jurisdictions in battleground states reported lower presidential drop-off than other jurisdictions, 1.03 versus 1.14 percent. Jurisdictions in battleground states reported lower Senate drop-off, 3.66 versus 9.19 percent, and lower U.S. House drop-off, versus percent, than other jurisdictions. Presidential Margin of Victory Jurisdictions with a higher margin of victory tended to report slightly higher presidential drop-off than those jurisdictions with a lower margin of victory, 0.96 versus 1.17 percent. There was no pattern and little variation in reported Senate drop-off among jurisdictions according to the presidential margin of victory, varying between 3.91 and 5.21 percent. There was no pattern and high variation in reported U.S. House drop-off among jurisdictions according to the presidential margin of victory. Red versus Blue Jurisdictions Similar to the margin of victory, jurisdictions that were won by the highest margin for either candidate tended to report higher levels of presidential drop-off, the same 1.21 percent for jurisdictions won overwhelmingly either by Bush or Kerry. Those where the election was closest reported the smallest drop-off, 0.96 and 0.78 respectively for jurisdictions won by Bush or Kerry by a plurality. Jurisdictions where the election was closest, where Bush or Kerry won by a plurality, reported the highest Senate drop-off, 9.20 and 5.85 percent respectively. All other jurisdictions varied between 4.03 and 5.12 percent. There was high variation in reported U.S. House drop-off among jurisdictions according to the presidential winner within the jurisdiction. Those jurisdictions won by a plurality by either candidate reported the highest levels of U.S. House drop-off, and percent respectively for jurisdictions won by Bush or Kerry by a plurality. REFERENCES Burnham, Walter Dean The Changing Shape of the American Political Universe. The American Political Science Review 59(1): Crocker, Royce Voter Registration and Turnout: CRS Report to Congress: CRS Washington, DC: Congressional Research Service. (This CRS report utilizes data compiled by Election Data Services, Inc.)

132 EAC Election Day Survey Drop-Off 2004 General Election Drop-Off Cases = Number of Jurisdictions Reporting Subject Matter Presidential U.S. Senate U.S. House of Representatives Updated: 09/19/ :04:40 Number Percent Number Percent Number Percent Election Total Not Voting Drop-off Total Pres. > Not Voting Drop-off Total Sen. > Not Voting Drop-off Total Cong. > Administration Ballots Total For For Total Ballots Total For For Total Ballots Total For For Total Ballots Code Name Jurisdictions Counted Cases President Cases President President Cases Exceptions Senate Cases Senate Senate Cases Exceptions Cong Dist Cases Cong Dist Cong Dist Cases Exceptions 01 Alabama 67 1,683, ,758, , ,751, , ,638, , Alaska 1 314, , , , , , , Arizona 15 2,038, ,013, , ,932, , ,869, , Arkansas 75 1,055, ,055, ,040, , , , California 58 12,359, ,266, , ,808, , ,265, ,155, Colorado 64 2,148, ,130, , ,107, , ,040, , Connecticut 169 1,595, ,578, , ,424, , ,428, , Delaware 3 377, , , , , District of Columbia 1 230, , , , , Florida 67 7,639, ,609, , ,429, , ,627, ,887, Georgia 159 3,317, ,304, , ,222, , ,256, ,060, Hawaii 5 431, , , , , , Idaho , , , , , , , Illinois 110 5,361, ,070, , ,998, , Indiana 92 2,512, ,467, , ,428, , ,866, , Iowa 99 1,513, ,488, , ,462, , ,431, , Kansas 105 1,199, ,188, , ,129, , ,156, , Kentucky 120 1,816, ,794, , ,724, , ,635, , Louisiana 64 1,956, ,943, , ,848, , ,035, , Maine , , , , , Maryland 24 2,395, ,386, , ,323, , ,228, , Massachusetts 351 2,927, ,912, , ,472, , Michigan 83 4,876, ,839, , ,628, , Minnesota 87 2,842, ,825, , ,721, , Mississippi 82 1,163, ,152, , ,116, , Missouri 116 2,765, ,731, , ,706, , ,749, , Montana , , , , , Nebraska , , , , , Nevada , , , , , , , New Hampshire , , , , , , , New Jersey 21 3,639, ,609, , ,284, , New Mexico , , , , , , New York 58 7,448, ,391, , ,702, , ,819, , North Carolina 100 3,571, ,501, , ,420, , ,409, , North Dakota , , , , , , , Ohio 88 5,730, ,627, , ,427, , Oklahoma 77 1,474, ,467, , ,455, , ,418, , Oregon 36 1,851, ,836, , ,780, , ,772, , Pennsylvania 67 3,006, Rhode Island , , , , , South Carolina 46 1,626, South Dakota , , , , , , , Tennessee 95 2,458, ,434, , Texas 254 7,507, ,410, , ,836, , Utah , , , , , , , Vermont , , , , , Virginia 134 3,223, ,198, , ,548, , Washington 39 2,885, ,859, , ,818, , ,729, , West Virginia , , , , , Wisconsin 1,910 3,009,491 1,880 2,992,340 1,897 45, , ,869,954 1, , , ,815,739 1, , , Wyoming , , , , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 1,990, ,990, ,959, , Virgin Islands 1 31, , , , , Total 6, ,862,353 6, ,968,736 6,290 1,160, , ,486,597 4,377 5,676, , ,338,384 6,039 11,669, , Maximum 1,910 12,359,633 1,880 12,266,320 1, , , ,808,639 1,897 1,990, , ,265,624 1,896 2,155, , Average 119 2,299, ,325, , ,242, , ,798, , Minimum 1 31, , , , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

133 EAC Election Day Survey Drop-Off 2004 General Election Drop-Off Cases = Number of Jurisdictions Reporting Subject Matter Presidential U.S. Senate U.S. House of Representatives Updated: 09/19/ :04:40 Number Percent Number Percent Number Percent Election Total Not Voting Drop-off Total Pres. > Not Voting Drop-off Total Sen. > Not Voting Drop-off Total Cong. > Administration Ballots Total For For Total Ballots Total For For Total Ballots Total For For Total Ballots Code Name Jurisdictions Counted Cases President Cases President President Cases Exceptions Senate Cases Senate Senate Cases Exceptions Cong Dist Cases Cong Dist Cong Dist Cases Exceptions Election Administration President, U.S. Senate, and U.S. House of Representatives Drop-Off excluded where Drop-Off = 0. Voting Equipment Used in 2004 General Election None / Unknown 908 9,922, ,782, , ,837, ,277, ,193, , Punch card ,938, ,202, , ,160, , ,618, , Lever ,981, ,663, , ,135, , ,439, ,334, Paper 1,734 2,172,234 1,727 2,135,810 1,732 29, , ,384,461 1,106 55, , ,464,574 1, , , Optical scan 2,541 49,661,061 2,524 48,141,235 2, , , ,045,332 1,589 1,125, , ,376,414 2,378 5,227, , Electronic ,295, ,285, , ,329, , ,563, ,402, Multiple Systems 123 8,891, ,758, , ,594, , ,682, ,008, Changed Voting Equipment Since 2000 General Election Yes 1,753 35,479,523 1,739 34,381,658 1, , , ,799,451 1,434 1,031, , ,546,492 1,629 5,195, ,622 8 No 4,815 86,382,830 4,749 79,587,078 4, , , ,687,146 2,943 4,645, , ,791,892 4,410 6,474, , State Wide Voter Registration System in Place Yes 1,335 27,317,939 1,322 25,489,478 1, , , ,740, , ,199,846 1,259 2,872, ,259 1 No 5,233 94,544,414 5,166 88,479,258 5, , , ,746,368 3,701 5,071, , ,138,538 4,780 8,797, , Election Day Registration Yes 2,823 8,152,145 2,792 8,077,465 2, , , ,030,935 2, , , ,711,699 2, , , No 3, ,710,208 3, ,891,271 3,484 1,055, , ,455,662 2,198 5,346, , ,626,685 3,232 11,201, , Provisional Ballot Acceptance In Overall Jurisdiction 1,162 44,662,901 1,123 41,055,629 1, , , ,842,876 1,004 1,491, , ,907, ,582, In Precinct Only 4,350 69,964,775 4,312 67,758,097 4, , , ,141,833 3,037 2,048, , ,472,533 4,077 6,808, , None 1,056 7,234,677 1,053 5,155, , ,501, ,136, ,957,861 1, , ,052 3 No Excuse Absentee Balloting Yes 3,781 46,531,514 3,731 46,108,172 3, , , ,268,719 3,041 1,797, , ,200,425 3,715 5,663, , No 2,787 75,330,839 2,757 67,860,564 2, , , ,217,878 1,336 3,878, , ,137,959 2,324 6,006, , Early Voting Allowed Yes 1,701 50,903,807 1,681 50,439,831 1, , , ,689,513 1,180 1,514, , ,451,444 1,541 6,674, , No 4,867 70,958,546 4,807 63,528,905 4, , , ,797,084 3,197 4,162, , ,886,940 4,498 4,994, , Covered By Section 203, Language Minority Requirements Yes ,287, ,986, , ,303, ,089, ,887, ,966, No 6,100 87,574,692 6,038 79,982,072 5, , , ,183,093 4,228 4,587, , ,450,641 5,592 6,702, , Covered By Section 5 of Voting Rights Act Yes ,429, ,638, , ,521, , ,772, ,473, No 5,688 94,432,928 5,616 88,330,456 5, , , ,965,170 4,016 5,106, , ,565,710 5,227 8,196, , Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

134 EAC Election Day Survey Drop-Off 2004 General Election Drop-Off Cases = Number of Jurisdictions Reporting Subject Matter Presidential U.S. Senate U.S. House of Representatives Updated: 09/19/ :04:40 Number Percent Number Percent Number Percent Election Total Not Voting Drop-off Total Pres. > Not Voting Drop-off Total Sen. > Not Voting Drop-off Total Cong. > Administration Ballots Total For For Total Ballots Total For For Total Ballots Total For For Total Ballots Code Name Jurisdictions Counted Cases President Cases President President Cases Exceptions Senate Cases Senate Senate Cases Exceptions Cong Dist Cases Cong Dist Cong Dist Cases Exceptions Demographics Region Northeast 1,710 20,812,375 1,687 17,662,003 1, , , ,098, , ,995,074 1,622 2,086, ,621 5 South 1,423 42,266,877 1,417 40,376,851 1, , , ,215, , ,841,229 1,229 5,367, , Midwest 2,902 31,316,030 2,871 30,710,408 2, , , ,724,317 2, , , ,836,204 2,673 1,350, , West ,445, ,219, , ,417, ,064, ,676, ,832, Territories 113 2,021, , ,991, ,989, , Urban to Rural Urban ,675, ,592, , ,244, ,502, ,198, ,123, Suburban ,263, ,993, , ,314, , ,915, ,118, Small Towns 1,710 30,364,561 1,685 28,304,035 1, , , ,398,374 1, , , ,603,090 1,461 1,682, , Rural 3,307 13,536,721 3,269 13,078,933 3, , , ,499,249 2, , , ,631,055 3, , , Not Available - Territories 113 2,021, , ,991, ,989, , Size of Jurisdiction (VAP) < 1,000 1, ,024 1, ,085 1,754 6, , ,044 1,425 25, , ,365 1,751 64, , >=1,000 to <3,500 1,165 1,630,543 1,162 1,597,029 1,164 40, , ,092, , ,502,245 1, , , >=3,500 to <10,000 1,043 4,256,986 1,038 4,156,692 1,029 60, , ,393, , ,742, , >=10,000 to <50,000 1,704 21,817,391 1,689 20,865,743 1, , , ,477,672 1, , , ,947,060 1,431 1,345, , >=50,000 to <250, ,587, ,211, , ,055, , ,517, ,077, >=250,000 to <1,000, ,485, ,329, , ,242, , ,788, ,043, >=1,000, ,427, ,168, , ,664, ,055, ,270, ,655, Not Available 144 2,022, , , ,991, ,992, , Race and Ethnicity Predominently NH White 6, ,362,361 6, ,734,482 6,127 1,055, , ,276,603 4,300 3,413, , ,272,192 5,767 10,597, , Predominently NH Black 85 2,117, ,960, , ,337, , ,377, , Predominently NH Native American , , , , , , , Predominently Hispanic 50 5,209, ,122, , ,729, , ,549, , Not Available 145 2,046, , , ,991, ,016, , Median Income < $25, ,488, ,440, , , , ,191, , >=$25,000 to <$30, ,685, ,455, , ,199, , ,234, , >=$30,000 to <$35,000 1,372 14,312,622 1,356 13,179,325 1, , , ,865, , ,170,815 1, , , >=$35,000 to <$40,000 1,215 26,144,458 1,204 24,868,484 1, , , ,552, , ,829,284 1,101 1,768, ,096 9 >=$40,000 to <$45, ,227, ,994, , ,228, , ,709, ,574, >=$45,000 to <$50, ,037, ,514, , ,040, , ,594, ,770, >=$50,000 1,180 30,943,728 1,173 29,512,042 1, , , ,759, , ,615,873 1,142 3,294, ,138 5 Not Available 151 2,022, , , ,991, ,992, , High School Education < 60% , , , , , , , >=60% to <70% ,083, ,719, , ,482, , ,292, , >=70% to <80% 1,646 31,406,406 1,630 29,496,147 1, , , ,778,364 1, , , ,677,376 1,469 2,034, , >=80% to <90% 3,111 65,275,468 3,074 61,770,600 3, , , ,838,970 2,242 2,035, , ,262,963 2,901 7,515, , >=90% ,099, ,023, , ,993, , ,311, , Not Available 151 2,046, , , ,991, ,016, , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

135 EAC Election Day Survey Drop-Off 2004 General Election Drop-Off Cases = Number of Jurisdictions Reporting Subject Matter Presidential U.S. Senate U.S. House of Representatives Updated: 09/19/ :04:40 Number Percent Number Percent Number Percent Election Total Not Voting Drop-off Total Pres. > Not Voting Drop-off Total Sen. > Not Voting Drop-off Total Cong. > Administration Ballots Total For For Total Ballots Total For For Total Ballots Total For For Total Ballots Code Name Jurisdictions Counted Cases President Cases President President Cases Exceptions Senate Cases Senate Senate Cases Exceptions Cong Dist Cases Cong Dist Cong Dist Cases Exceptions Political Battleground States in 2004 Presidential Election Yes 3,093 43,980,255 3,028 40,593,153 2, , , ,048,325 2,750 1,164, , ,660,529 2,870 3,396, , No 3,475 77,882,098 3,460 73,375,583 3, , , ,438,272 1,627 4,512, , ,677,855 3,169 8,273, , Margin of Victory in 2004 Presidential Election < 2.5% ,753, ,520, , ,307, , ,345, ,257, >=2.5% to < 5.0% 476 8,077, ,840, , ,912, , ,877, , >=5.0% to < 7.5% 510 9,931, ,790, , ,359, , ,979, ,145, >=7.5% to < 10.0 % 429 6,126, ,948, , ,412, , ,322, , >=10.0 % 4,492 84,945,042 4,448 80,860,548 4, , , ,455,145 3,044 2,605, , ,816,606 4,117 7,531, , Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 47,293,906 3,083 44,578,904 3, , , ,588,738 2,176 1,215, , ,235,076 2,814 3,469, , Bush 50% to 55% ,343, ,379, , ,240, , ,785, ,067, Bush < 50% 136 1,386, ,318, , ,114, , ,008, , Kerry < 50% 150 3,447, ,423, , ,476, , ,418, , Kerry 50% to 55% ,109, ,319, , ,167, , ,476, ,887, Kerry > 55% 1,161 33,249,808 1,152 31,936,516 1, , , ,860, ,256, ,413,183 1,102 3,722, , Tied 25 9, , , , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

136 Chapter 8 Overvotes and Undervotes Table 8 presents data from the Election Day Survey on overvotes and undervotes for candidates in federal contests at the November 2004 general election. Traditionally, overvotes and undervotes are the two components that combine together to make up drop-off. (See chapter 7). While drop-off is a data element calculated from known data items, overvotes and undervotes are actual numbers that have to be generated by vote-tallying software. The U.S. Election Assistance Commission (EAC) survey defined overvote as an occurrence when a voter makes more than the permitted number of selections in a single race/contest or when a voter makes a selection in a race/contest on which he/she was not eligible to vote. The problems with the EAC definition are how to operationally test for the latter part of the definition and be able to collect that data. Traditionally, overvotes occur when a voter casts more votes for an office than they are allowed to cast. For example, in a vote for one office like president, a voter casting votes for two or more candidates would have overvoted for that office only. As a result, the voter s candidate choice would not be recorded for that office. The voter s choices for other offices, if properly cast, would be counted under most state laws. Independent research has shown that the higher the allowable votes for an office (like a vote for five contest), the higher the rate of overvotes. It would appear that voters have a hard time keeping track of how many candidates they have voted for. Overvoting usually comes about because of voter mistakes and/or bad ballot design. Actual data has shown that when candidates for a single office are spread across multiple pages in a ballot book, or multiple columns on a ballot, greater numbers of overvotes occur. Many times voters fall prey to mistakes they have no control over. The EAC defined an undervote as an occurrence when a voter makes less than that allowed number of selections in a single race/contest or when a voter votes on less than all of the races/contests for which he/she is eligible to vote. It is possible that an undervote can occur when a voter simply does not see a contest, and then fails to vote for any candidate for that office. But it is also possible, in fact, more likely, that an undervote occurs through the voter s choice to not cast a vote for that office. This choice may be because they do not know anything about the candidates and do not want to make an uneducated choice, or they uniformly do not like any of the candidates. It is not proper to attribute all of the undervotes to errors in voting equipment, as some commentators have subscribed, but it is possible that a small number may be linked to faulty voting equipment. Applicability and Coverage For some voting systems, such as electronic direct recording electronic (DRE) machines, overvotes can be prevented through programming in the units. As a result, overvote reports tend to not be

137 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-2 September 27, 2005 generated in jurisdictions using that type of system. The EAC survey shows that only one-third of the nation s jurisdictions reported the number of overvotes. Overvote and undervote reports have traditionally been difficult to obtain. The EAC survey was able to obtain undervote counts from just two-thirds of the nation s jurisdictions. Historical Context Overvotes and undervotes traditionally have been tied to advances and variation in voting technology. When votes were cast by voice or by party-printed ballots in the 19th century, overvotes and undervotes would have been more difficult to cast. A person voicing their vote for multiple candidates when only one was allowed would have been corrected on the spot, and abstention would have been specifically voiced by the voter. Similarly, with party-printed ballots, voters would have had to alter the existing ballot to cast an overvote or an undervote. In 1888, Massachusetts became the first state to adopt the secret ballot (Evans 1917), which moved administration of printing ballots away from the political parties and to local jurisdictions. As more jurisdictions adopted the secret ballot, many maintained the use of paper ballots, but provided boxes for voters to check off the candidate they wanted to vote for. An early alternative system of lever machines was adopted in New York in Voters experiences varied with the format of the ballot: some jurisdictions created office bloc formats where voters selected among candidates for offices, while other adopted party bloc formats that allowed the selection of all candidates for one party with one check of a box. While the party bloc format reduced the amount of split-ticket voting (Rusk 1970), it also likely reduced the incidence of overvotes and undervotes for offices on the ballot (Walker 1966). There are few historical statistics on the incidence of overvoting and undervoting. Research specific to overvoting and undervoting, rather than the broader subject of drop-off or residual votes, first examined municipal elections where multiple candidates could be selected for a single office, such as school board, and compared rates of overvoting and undervoting using paper and punch card ballots in an experimental setting (Shocket, Heighberger, and Brown 1992). In the wake of the 2000 election, a number of academic studies (e.g., Brady, Herron, Mebane, Sekhon, Shotts, and Wand 2001) and a consortium of news organizations (Keating and Balz 2001) examined the structure of the Florida ballot in relation to overvoting and undervoting for president. Traditionally (i.e., before 2000), vote-tallying software did not report undervotes and overvotes unless specifically programmed to generate such numbers. Vote equipment and vote-tallying software vendors usually have not advocated to election administrators that they exercise any option to print overvotes and undervotes with the election results. This study s principal investigator has over three decades of observing elections and studying election results, but has found only a few instances of overvote and undervote tallies. Where we have found such data, generally the information is heavily tilted toward the undervotes, with few overvotes recorded. It is typical to find that only about 10 percent of the drop-off in a contest is caused by overvoting, while 90 percent of the drop-off is due to undervoting.

138 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-3 September 27, 2005 Survey Results Table 8 presents data on overvotes and undervotes from question 12 on the Election Day Survey. In the table, the overvotes and undervotes, which are the components of drop-off, are calculated for presidential, senatorial, and congressional elections as percentages of ballots counted. The column headings in Table 8 are as follows: Table 8 Column Headings. Overvotes and undervotes Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Ballots Counted Number of ballots counted from survey question 2 5 Cases Number of jurisdictions that responded to question 2 6 Total Ballots for President Number of votes for president from survey question 12 7 Cases Number of jurisdictions that responded to question 12 8 Number Not Voting for President (Drop-off) 9 Percent Not Voting for President (Drop-off) Number of ballots counted (col. 4) minus the number of votes for president (col. 6) Number not voting for president (col. 8) divided by the number of ballots counted (col. 4). 10 Cases Number of jurisdictions that responded to questions 2 and Exceptions Number of jurisdictions that reported more votes for president (col. 6) than the total number of ballots counted (col. 4) 12 Total Overvotes for President Number of overvotes for president from survey question Cases Number of jurisdictions that responded to question Percent President Overvotes of Total Ballots Number of overvotes for president (col. 12) divided by the number of ballots counted (col. 4). 15 Cases Number of jurisdictions that responded to questions 2 and Percent Pres. Overvotes of Total Over & Undervotes Number of overvotes for president (col. 14) divided by the sum of the number of overvotes for president (col. 12) and the number of undervotes for president (col. 18). 17 Cases Number of jurisdictions that responded to questions 10, 11, and Total Undervotes for President Number of undervotes for resident from survey question Cases Number of jurisdictions that responded to question Percent President Undervotes of Total Ballots Number of undervotes for president (col. 18) divided by the number of ballots counted (col. 4). 21 Cases Number of jurisdictions that responded to questions 2 and 10

139 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-4 September 27, 2005 Table 8 Column Headings (cont.) Col. Heading Description 22 Percent Pres. Undervotes of Total Over & Undervotes Number of undervotes for president (col. 18) divided by the sum of the number of overvotes for president (col. 12) and the number of undervotes for president (col. 18). 23 Cases Number of jurisdictions that responded to questions 10, 11, and Total Ballots for U.S. Senate Number of votes for U.S. Senator from survey question Cases Number of jurisdictions that responded to question Number Not Voting for U.S. Senate (Drop-off) 27 Percent Not Voting for U.S. Senate (Drop-off) Number of ballots counted (col. 4) minus the number of votes for U.S. Senator (col. 24) Number not voting for U.S. Senator (col. 26) divided by the number of ballots counted (col. 4). 28 Cases Number of jurisdictions that responded to questions 2 and Exceptions Number of jurisdictions that reported more votes for U.S. Senator (col. 24) than the total number of ballots counted (col. 4) 30 Total Overvotes for U.S. Senate Number of overvotes for U.S. Senator from survey question Cases Number of jurisdictions that responded to question Percent U.S. Senate Overvotes of Total Ballots Number of overvotes for U.S. Senator (col. 30) divided by the number of ballots counted (col. 4). 33 Cases Number of jurisdictions that responded to questions 2 and Percent U.S. Sen. Overvotes of Total Over & Undervotes Number of overvotes for U.S. Senator (col. 30) divided by the sum of the number of overvotes for U.S. Senator (col. 30) and the number of undervotes for U.S. Senator (col. 36). 35 Cases Number of jurisdictions that responded to questions 10, 11, and Total Undervotes for U.S. Senate Number of undervotes for U.S. Senator from survey question Cases Number of jurisdictions that responded to question Percent U.S. Senate Undervotes of Total Ballots Number of undervotes for U.S. Senator (col. 36) divided by the number of ballots counted (col. 4). 39 Cases Number of jurisdictions that responded to questions 2 and Percent U.S. Sen. Undervotes of Total Over & Undervotes Number of undervotes for U.S. Senator (col. 36) divided by the sum of the number of overvotes for U.S. Senator (col. 30) and the number of undervotes for U.S. Senator (col. 36). 41 Cases Number of jurisdictions that responded to questions 10, 11, and Total Ballots for Congress Number of votes for U.S. Representative from survey question Cases Number of jurisdictions that responded to question 12

140 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-5 September 27, 2005 Table 8 Column Headings (cont.) Col. Heading Description 44 Number Not Voting for Congress (Drop-off) 45 Percent Not Voting for Congress (Drop-off) Number of ballots counted (col. 4) minus the number of votes for U.S. Representative (col. 42) Number not voting for U.S. Representative (col. 44) divided by the number of ballots counted (col. 4). 46 Cases Number of jurisdictions that responded to questions 2 and Exceptions Number of jurisdictions that reported more votes for U.S. Representative (col. 42) than the total number of ballots counted (col. 4) 48 Total Overvotes for Congress Number of overvotes for U.S. Representative from survey question Cases Number of jurisdictions that responded to question Percent Congress Overvotes of Total Ballots This cell has an extra line of space at the top. Number of overvotes for U.S. Representative (col. 48) divided by the number of ballots counted (col. 4). 51 Cases Number of jurisdictions that responded to both question 2 and Percent Congress Overvotes of Total Over & Undervotes Number of overvotes for U.S. Representative (col. 48) divided by the sum of the number of overvotes for U.S. Representative (col. 48) and the number of undervotes for U.S. Representative (col. 54). 53 Cases Number of jurisdictions that responded to questions 10, 11, and Total Undervotes for Congress Number of undervotes for U.S. Representative from survey question Cases Number of jurisdictions that responded to question Percent Congress Undervotes of Total Ballots Number of undervotes for U.S. Representative (col. 54) divided by the number of ballots counted (col. 4). 57 Cases Number of jurisdictions that responded to both questions 2 and This cell has an extra line of space at the top. Percent Undervotes of Total Over & Undervotes Number of undervotes for U.S. Representative (col. 54) divided by the sum of the number of overvotes for U.S. Representative (col. 48) and the number of undervotes for U.S. Representative (col. 54). 59 Cases Number of jurisdictions that responded to questions 10, 11, and 12

141 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-6 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 8 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 8 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary Drop-off, which is a combination of undervotes and overvotes, is analyzed in chapter 7, and although a column is reported for presidential, Senate, and U.S. House drop-off, it is not discussed in depth in this chapter. Nationally, for president, 133,289 overvotes were reported cast, or 0.23 percent of total ballots cast. 863,872 presidential undervotes were reported, or 0.91 percent of total ballots cast. For Senate, 49,100 overvotes were reported, or 0.11 percent of total ballots cast in Senate contests. For Senate races, 2,488,016 Senate undervotes were reported, or 3.80 percent of total ballots cast. For U.S. House, 56,173 overvotes were reported, or 0.12 percent of total ballots cast in House contests. 5,077,325 undervotes in House elections were reported, or 6.27 percent of total ballots cast. Considerable care should be taken in interpreting the analysis presented here. Many jurisdictions did not provide overvotes and undervotes on their response to the Election Day Survey, so the analysis is only valid for reporting jurisdictions. In addition, some jurisdictions provided data for either overvotes or undervotes, but not both. We also make two caveats for overvotes and undervotes: For overvotes, only a small percentage of overvotes were reported cast, and any inference is suspect when there is little variation. For undervotes, particularly for Senate and U.S. House, undervotes are a function of the competitiveness of the election. We note sizable increases in undervotes for elections that were won handily by one candidate, particularly when that candidate was unopposed.

142 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-7 September 27, 2005 With these caveats in mind, we note a few interesting patterns in our analysis: Jurisdictions with the lowest income and education levels tended to report the highest percentage of overvotes and undervotes. The percentages tended to drop to a lower level at the second-to-third lowest income and education categories. Many of the studies in the wake of the 2000 election focused on the relationship between voting equipment and overvotes and undervotes. For overvotes, which are more clearly an error in the recording of a vote, we note that punch card jurisdictions reported the highest overvotes as a percentage of total ballots cast for president and Senate, and were second highest to paper ballot jurisdictions for U.S. House elections. Overvotes are more prone to occur in voting systems where a voter marks a physical ballot and it is deposited in a ballot box to be counted at the close of the polls. Systems that provide an in-precinct checking capability are likely to see lower numbers of overvotes. Electronic systems reported a low percentage of undervotes, but lever systems also reported a low rate, as did jurisdictions using multiple systems. Optical scan jurisdictions tended to fall in the middle. Predominantly Hispanic jurisdictions tended to report the highest percentage of overvotes for all offices, and generally a high percentage of undervotes for U.S. House and Senate. Predominantly non-hispanic Native American jurisdictions reported the highest percentage of undervotes for president. States For the presidential election, Alaska reported the lowest overvotes as a percentage of total ballots cast, 0.01 percent. Idaho reported the highest, 1.50 percent. Alabama and Maryland reported the lowest undervotes as a percentage of total ballots cast, 0.30 percent. Nevada also reported a low number of undervotes, 0.44 percent; Nevada is the only state that presents voters with None of these candidates as an option for presidential vote. New Mexico reported the highest percentage of undervotes, 2.74 percent, but this was based on reports of only 10 of the state s 33 counties. For Senate, Maryland reported the lowest overvotes as a percentage of total ballots cast, percent. Illinois reported the highest percent of overvotes at 0.27 percent. Colorado reported the lowest percentage of undervotes at 0.47 percent and Idaho reported the highest, at percent. The high percentage in Idaho and the second highest in Connecticut at percent, reflect the lack of competition for the office of Senate in these states both winning candidates won handily. Democrats in Idaho chose not to nominate a candidate, though a Democratic write-in did receive a small number of votes. For U.S. House, Maryland reported the lowest undervotes as a percentage of total ballots cast, percent, and Washington reported the highest, at 0.84 percent. Puerto Rico reported the lowest number of undervotes for its nonvoting member of Congress, 0.25 percent, and North Dakota was next, at 1.51 percent. Massachusetts and Connecticut reported undervotes for U.S. House slightly higher than 10 percent.

143 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-8 September 27, 2005 Regions For president, reported percentage overvotes were lowest in the Northeast and South, 0.03 and 0.08 percent, respectively, and were greatest in the Midwest and West, 0.36 and 0.25 percent, respectively. Percentage undervotes were generally equal across regions, with the Midwest reporting the highest, 1.00 percent, and the Northeast the lowest, 0.75 percent. For Senate, reported percentage overvotes were lowest in the Northeast and South, 0.02 and 0.01 percent, respectively, and were greatest in the Midwest and West, 0.17 and 0.11 percent, respectively. Percentage undervotes were greatest in the Northeast, 9.29 percent, and relatively similar elsewhere, varying between 3.12 and 3.99 percent. For U.S. House, reported percentage overvotes was greatest in the U.S. territories, 0.20 percent, followed by the West, 0.19 percent; Midwest, 0.12 percent; and Northeast, 0.10 percent. The South reported the lowest percentage at 0.02 percent. Percentage of undervotes was greatest in the Northeast, 9.76 percent, followed by the South, 6.59 percent, and the West and Midwest, which were 5.97 and 4.49 percent, respectively. The U.S. territories reported percentage of undervotes at 0.25 percent. Urban to Rural For president, urban and suburban jurisdictions reported the lowest percentage of overvotes, 0.15 and 0.13 percent, respectively. Small towns and rural jurisdictions reported higher percentages, 0.18 and 0.33 percent, respectively. Percentage of undervotes was highest among rural jurisdictions, 1.34 percent, and similar elsewhere, 0.77 to 0.88 percent. For Senate, urban jurisdictions reported the highest percentage of overvotes, 0.12 percent, and other jurisdictions varied between 0.04 and 0.07 percent. Percentage of undervotes was also highest in urban areas, 4.39 percent, and varied between 3.17 and 3.71 percent elsewhere. For U.S. House, suburban jurisdictions reported 0.21 percent overvotes, while other jurisdictions reported 0.05 to 0.09 percent. Urban jurisdictions reported the highest percentage of undervotes, 7.43 percent, and jurisdictions elsewhere reported around 6.00 percent. The territories reported overvotes of 0.20 percent and undervotes of 0.25 percent. Size of Jurisdiction For president, jurisdictions with voting age population (VAP) in the two categories 3,500 to 10,000 and 10,000 to 50,000 reported the highest percentage of overvotes, 0.31 and 0.28 percent, respectively. All other jurisdictions reported between 0.12 and 0.18 percent. Jurisdictions with VAP between 1,000 to 3,500 reported the highest percentage of undervotes, 2.66 percent, followed by 10,000 to 50,000, 1.29 percent; and 3,500 to 10,000, 1.15 percent; followed by all other jurisdictions at 0.62 to 1.03 percent. For Senate, reported overvotes ranged as a percentage between 0.05 and 0.12 percent, with no clear pattern. Percentage undervotes tended to be higher in smaller jurisdictions, those with populations of up to 50,000 voting-age population reporting over 4.43 percent and larger jurisdictions all reporting below 3.89 percent.

144 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-9 September 27, 2005 For U.S. House, smaller jurisdictions under 3,500 reported higher percentages of overvotes and undervotes than larger jurisdictions. The smallest jurisdictions, under 1,000 voting age population, reported the highest percentage of overvotes and undervotes, 0.32 and percent, respectively. Race and Ethnicity For president, predominantly Hispanic jurisdictions reported much higher percentages of overvotes, 0.29 percent, than for predominantly non-hispanic White and predominantly non-hispanic Black jurisdictions, 0.17 and 0.10 percent, respectively. Predominantly non-hispanic Native American and predominantly Hispanic jurisdictions reported higher percentages of undervotes, 3.93 and 1.40 percent, respectively, than for predominantly non-hispanic White and predominantly non-hispanic Black jurisdictions, 0.87 and 0.57 percent, respectively. For Senate, predominantly Hispanic jurisdictions reported a higher percentage of overvotes, 0.19 percent, than all other jurisdictions, which reported 0.06 percent and below. Predominantly Hispanic reported higher percentages of undervotes, 5.22 percent, than other jurisdictions, which reported between 3.34 and 3.93 percent. For U.S. House, predominantly Hispanic and predominantly non-hispanic White jurisdictions reported higher percentages of overvotes, 0.12 percent for both, than for predominantly non- Hispanic Native American, at 0.08 percent, and predominantly non-hispanic Black jurisdictions, at 0.02 percent. Predominantly Hispanic jurisdictions reported the highest percentage of undervotes, 8.96 percent, and predominantly non-hispanic Native American jurisdictions reported the lowest, at 4.44 percent, while all other jurisdictions varied between 6.15 and 6.26 percent. Median Income For president, reported percentage of overvotes tended to decrease as median income within a jurisdiction increased, from 0.31 to 0.08 percent. The same pattern is evident in percentage of undervotes, which tended to decrease as median income in a jurisdiction increased, from 1.83 to 0.65 percent. For Senate, reported percentage of overvotes showed no clear pattern, varying between 0.04 and 0.09 percent. For percentage of undervotes, the lowest median income jurisdiction reported the highest percent, 5.31, but varied between 3.51 and 3.92 for the remaining jurisdictions without exhibiting a clear pattern. For U.S. House, for percentage of overvotes, the second to highest [ second highest?] median income jurisdictions reported the highest level, 0.22 percent, and the highest median income jurisdictions the next highest level, 0.16 percent, while the remaining jurisdictions measured between 0.05 and 0.12 percent. For percentage of undervotes, the lowest median income jurisdiction reported the highest level, 8.38 percent,. The remaining jurisdictions varied between 5.37 and 7.44 percent without exhibiting a clear pattern. High School Education For president, reported percentage of overvotes was highest in jurisdictions for the two lowest categories of education, 0.22 percent for the lowest, 0.34 percent for the second lowest, and 0.16

145 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-10 September 27, 2005 percent or lower for the remainder. Reported percentage of undervotes trended down with increasing education, ranging from 1.66 to 0.48 percent. For Senate, jurisdictions in the second lowest and third lowest categories of education reported the highest percentage of overvotes, 0.15 and 0.08 percent, respectively. The lowest category reported 0.03 percent, the second highest reported 0.06 percent and the highest reported 0.03 percent. Jurisdictions with the two lowest education categories reported the highest percentage of undervotes, 5.64 percent for the lowest and 4.65 percent for the second lowest. The remainder varied between 3.34 and 3.95 percent. For U.S. House, there was no clear pattern to reported overvotes, which ranged between 0.07 and 0.26 percent. Percentage of undervotes tended to follow a pattern of decreasing overvotes with increasing education, though jurisdictions in the second lowest education category reported the highest percentage of undervotes, 8.93 percent. The highest education category reported the lowest percentage, 5.50 percent. Section 203 Language Minority Requirements For president, Section 203 covered jurisdictions reported the same percentage of overvotes as other jurisdictions, 0.17 percent, and reported a higher percentage of undervotes than other jurisdictions, 1.04 versus 0.83 percent. For Senate, Section 203 covered jurisdictions reported a higher percentage of overvotes than other jurisdictions, 0.10 versus 0.05 percent, and reported a slightly lower percentage of undervotes than other jurisdictions, 3.73 versus 3.81 percent. For U.S. House, Section 203 covered jurisdictions reported a slightly lower percentage of overvotes than other jurisdictions, 0.11 versus 0.13 percent, and reported a higher percentage of undervotes than other jurisdictions, 7.02 versus 5.88 percent. Section 5 Preclearance of Voting Procedures For president, Section 5 covered jurisdictions reported a lower percentage of overvotes than other jurisdictions, 0.12 versus 0.20 percent, and reported a higher percentage of undervotes than other jurisdictions, 1.08 versus 0.84 percent. For Senate, Section 5 covered jurisdictions reported a slightly lower percentage of overvotes than other jurisdictions, 0.06 versus 0.08 percent, and reported a slightly higher percentage of undervotes than other jurisdictions, 3.87 versus 3.76 percent. For U.S. House, Section 5 covered jurisdictions reported a lower percentage of overvotes than other jurisdictions, 0.08 versus 0.14 percent, and reported a higher percentage of undervotes than other jurisdictions, 7.37 versus 5.84 percent. Type of Voting Equipment For president, jurisdictions with lever machines reported the lowest percentage of overvotes, percent. Electronic machines were next lowest at 0.03 percent. Although electronic machines inform voters that an overvote is an error, voters in these jurisdictions can still cast absentee and provisional ballots that may produce overvotes. Multiple system jurisdictions reported 0.06 percent overvotes. Punch card jurisdictions reported the highest percentage of overvotes, 0.49 percent, and paper and optical scan jurisdictions reported 0.22 and 0.21 percent overvotes, respectively. Punch card jurisdictions also reported the highest percentage of undervotes, 1.41 percent, followed by paper at

146 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-11 September 27, percent, lever at 0.94 percent, optical scan at 0.86 percent, and electronic and multiple systems, both at 0.70 percent. For Senate, electronic and multiple system jurisdictions reported the lowest percentage of overvotes at 0.02 percent, followed by, in increasing order, paper, 0.06 percent; optical scan, 0.08 percent; and punch card, 0.24 percent. (No jurisdictions with lever machines reported overvotes in a Senate race.) For undervotes, lever jurisdictions reported the highest percentage of undervotes, 9.17 percent, followed by punch cards, 4.08 percent;, electronic, 3.60 percent; paper, 3.33 percent; optical scan, 3.27 percent; and multiple systems, 2.23 percent. For U.S. House, paper jurisdictions reported the highest percentage of overvotes, 0.35 percent, and in descending order, punch cards at 0.22 percent, optical scan at 0.15 percent, lever and electronic at 0.05 percent, and multiple systems at 0.02 percent. Lever jurisdictions reported the highest percentage of undervotes, at percent, and in descending order, paper at 9.77 percent, electronic at 7.10 percent, optical scan at 5.88 percent, punch card at 5.01 percent, and multiple systems at 4.71 percent. Changed Voting Equipment since 2000 For all three types of federal offices, jurisdictions that changed voting equipment reported a lower percentage of overvotes and undervotes than jurisdictions that did not change voting equipment. Statewide Voter Registration Database For president, jurisdictions with a statewide voter registration database reported a lower percentage of overvotes than other jurisdictions, 0.08 versus 0.19 percent, and a lower percentage of undervotes than other jurisdictions, 0.73 versus 0.95 percent. For Senate, jurisdictions with a statewide voter registration database reported a lower percentage of overvotes than other jurisdictions, 0.03 versus 0.08 percent, and reported a higher percentage of undervotes than other jurisdictions, 4.40 versus 3.65 percent. For U.S. House, jurisdictions with a statewide voter registration database reported a lower percentage of overvotes than other jurisdictions, 0.08 versus 0.13 percent, and reported a higher percentage of undervotes than other jurisdictions, 6.95 versus 6.04 percent. Election Day Registration For president, jurisdictions with Election Day registration reported a higher percentage of overvotes than other jurisdictions, 1.27 versus 0.16 percent, and reported a higher percentage of undervotes than other jurisdictions, 1.08 versus 0.88 percent. For Senate, jurisdictions with Election Day registration reported a lower percentage of overvotes than other jurisdictions, 0.02 versus 0.07 percent, and reported a higher percentage of undervotes than other jurisdictions, 7.89 versus 3.55 percent. For U.S. House, jurisdictions with Election Day voter registration reported a slightly higher percentage of overvotes than other jurisdictions, 0.14 versus 0.12 percent, and reported a lower percentage of undervotes than other jurisdictions, 5.65 versus 6.33 percent. Provisional Ballot Acceptance For president, jurisdictions with no provisional ballots reported a higher percentage of overvotes, 1.02 percent, than other jurisdictions, which reported 0.18 percent of overvotes for within-

147 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-12 September 27, 2005 jurisdiction acceptance and 0.14 percent for within-precinct acceptance. Percentage of undervotes did not vary greatly by jurisdiction with regard to provisional ballot acceptance, ranging from 0.66 to 0.91 percent. For Senate, jurisdictions with provisional ballots accepted in the overall jurisdiction reported the highest rate of overvotes, 0.11 percent, versus 0.03 percent for jurisdictions with within-precinct acceptance and no provisional ballots. Jurisdictions without provisional ballots had the highest percentage of undervotes, percent (mainly due to Idaho[-why is that again?]), while other jurisdictions reported similar percentages: 3.54 percent in jurisdictionwide acceptance and 3.85 percent within-precinct acceptance. For U.S. House, jurisdictions with no provisional ballots and within jurisdiction acceptance of ballots reported the highest percentage of overvotes, 0.18 percent, versus 0.04 percent for withinprecinct acceptance. Jurisdiction with no provisional ballots reported the lowest percentage of undervotes, 2.67 percent, with 6.02 percent for jurisdictionwide acceptance and 6.94 for withinprecinct acceptance. No Excuse Absentee Balloting For president, jurisdictions with no excuse absentee balloting reported a slightly higher percentage of overvotes than other jurisdictions, 0.18 versus 0.17 percent, and reported a slightly lower percentage of overvotes than other jurisdictions, 0.88 versus 0.91 percent. For Senate, jurisdictions with no excuse absentee balloting reported a slightly lower percentage of overvotes than other jurisdictions, 0.07 versus 0.08 percent, and reported a slightly lower percentage of undervotes than other jurisdictions, 3.77 versus 3.80 percent. For U.S. House, jurisdictions with no excuse absentee balloting reported a higher percentage of overvotes than other jurisdictions, 0.14 versus 0.07 percent, and reported a lower percentage of undervotes than other jurisdictions, 6.04 versus 6.52 percent. Early Voting For president, jurisdictions with early voting reported a slightly lower percentage of overvotes than other jurisdictions, 0.17 versus 0.18 percent, and reported a higher percentage of overvotes than other jurisdictions, 0.95 versus 0.85 percent. For Senate, jurisdictions with early voting reported a slightly lower percentage of overvotes than other jurisdictions, 0.06 versus 0.10 percent, and reported a lower percentage of undervotes than other jurisdictions, 3.54 versus 4.09 percent. For U.S. House, jurisdictions with early voting reported a lower percentage of overvotes than other jurisdictions, 0.06 versus 0.26 percent, and reported slightly a higher percentage of undervotes than other jurisdictions, 6.35 versus 6.18 percent. Battleground States For president, jurisdictions in battleground states reported a slightly higher percentage of overvotes than other jurisdictions, 0.23 versus 0.15 percent, and reported a lower percentage of undervotes than other jurisdictions, 0.80 versus 0.96 percent. For Senate, jurisdictions in battleground states reported a slightly lower percentage of overvotes than other jurisdictions, 0.07 versus 0.08 percent, and reported a lower percentage of undervotes than other jurisdictions, 3.29 versus 4.15 percent. For U.S. House, jurisdictions in battleground states reported a higher percentage of overvotes than other

148 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-13 September 27, 2005 jurisdictions, 0.19 versus 0.09 percent, and reported a lower percentage of undervotes than other jurisdictions, 5.64 versus 6.59 percent. Presidential Margin of Victory For president, there was no clear pattern to reported percentage of overvotes by presidential margin of victory within the jurisdiction, ranging from 0.10 to 0.18 percent. There was no clear pattern for undervoters, either, ranging from 0.80 to 0.91 percent across categories. For Senate, there was no clear pattern to reported percentage of overvotes by presidential margin of victory within the jurisdiction, ranging from 0.03 to 0.07 percent. Reported percentage of undervotes also did not exhibit a clear pattern with presidential margin of victory within the jurisdiction, ranging from 3.02 to 4.52 percent. For U.S. House, the percentage of overvotes tended to decrease with presidential margin of victory within the jurisdiction, ranging from 0.37 to 0.02 percent. Reported percentage of undervotes tended to increase slightly with presidential margin of victory within the jurisdiction, ranging from 5.60 to 6.60 percent. Red versus Blue Jurisdictions For president, there was no clear pattern to overvotes by the presidential winner within the jurisdiction, ranging from 0.7 to 0.20 percent across categories. Jurisdictions won by Bush reported higher undervotes, ranging between 0.90 and 0.93 percent, than jurisdictions won by Kerry, which ranged from 0.77 to 0.86 percent. For Senate, those jurisdictions won by Kerry tended to report a higher percentage of overvotes, 0.05 to 0.11 percent, than jurisdictions won by Bush, ranging from 0.01 to 0.05 percent. Jurisdictions that Kerry won by a plurality reported the highest percentage of undervotes, 9.37 percent, while the remainder varied between 3.34 and 4.27 percent. (The few tied jurisdictions reported the highest undervote, 6.36 percent.) For U.S. House, those jurisdictions won by Bush tended to report a lower percentage of overvotes, ranging between 0.01 and 0.08 percent, than those won by Kerry, ranging between 0.08 and 0.21 percent. (The few tied jurisdictions reported an overvote of 0.17 percent.) There was no clear pattern for undervotes, which varied between 5.20 percent in jurisdictions won by Kerry by a plurality, to 7.90 percent for those jurisdictions won by Bush by a plurality. (The few tied jurisdictions reported the highest undervote, percent.) REFERENCES Brady, Henry E., Michael C. Herron, Walter R. Mebane Jr., Jasjeet Singh Sekhon, Kenneth W. Shotts, and Jonathan Wand Law and Data : The Butterfly Ballot Episode. PS: Political Science and Politics 34(1):

149 2004 Election Day Survey Report, Part 2 Survey Results Overvotes and undervotes, Page 8-14 September 27, 2005 Evans, Eldon C A History of the Australian Ballot System in the United States. Chicago: University of Chicago Press. Keating, Dan and Dan Balz Florida Recounts Would Have Favored Bush: But Study Finds Gore Might Have Won Statewide Tally of All Uncounted Ballots. The Washington Post, November 12, 2001: A1. Rusk, Jerrold G The Effect of the Australian Ballot Reform on Split Ticket Voting, American Political Science Review 64: Shocket, Peter A., Neil R. Heighberger, and Clyde Brown The Effect of Voting Technology on Voting Behavior in a Simulated Multi-Candidate City Council Election: A Political Experiment of Ballot Transparency. The Western Political Quarterly 45(2): Walker, Jack L Ballot Forms and Voter Fatigue: An Analysis of the Office Block and Party Column Ballots. Midwest Journal of Political Science 10:

150 EAC Election Day Survey Over and Undervotes 2004 General Election Overvotes and Undervotes Presidential Drop-off Presidential Overvotes Presidential Undervotes Updated: 09/19/ :05:59 Total Number Percent Total Percent Percent Total Percent Percent Election Total Ballots Not Voting Not Voting Overvotes President Overvotes of Undervotes President Undervotes of Administration Ballots For For President For President For OverVotes of Total Over & For Undervotes of Total Over & Code Name Jurisdictions Counted Cases President Cases (Drop-off) (Drop-off) Cases Exception President Cases Total Ballots Cases Undervotes Cases President Cases Total Ballots Cases Undervotes Cases 01 Alabama 67 1,683, ,758, , , Alaska 1 314, , , Arizona 15 2,038, ,013, , , , Arkansas 75 1,055, ,055, , , California 58 12,359, ,266, , , , Colorado 64 2,148, ,130, , , , Connecticut 169 1,595, ,578, , , Delaware 3 377, , , , District of Columbia 1 230, , , , Florida 67 7,639, ,609, , , , Georgia 159 3,317, ,304, , , Hawaii 5 431, , , Idaho , , , , , Illinois 110 5,361, ,070, , , , Indiana 92 2,512, ,467, , Iowa 99 1,513, ,488, , , , Kansas 105 1,199, ,188, , , Kentucky 120 1,816, ,794, , Louisiana 64 1,956, ,943, , , Maine , , , Maryland 24 2,395, ,386, , , Massachusetts 351 2,927, ,912, , , Michigan 83 4,876, ,839, , , Minnesota 87 2,842, ,825, , , Mississippi 82 1,163, ,152, , Missouri 116 2,765, ,731, , , Montana , , , , , Nebraska , , , , , Nevada , , , , New Hampshire , , , New Jersey 21 3,639, ,609, , , New Mexico , , , , New York 58 7,448, ,391, , North Carolina 100 3,571, ,501, , , North Dakota , , , , Ohio 88 5,730, ,627, , , , Oklahoma 77 1,474, ,467, , , , Oregon 36 1,851, ,836, , , , Pennsylvania 67 3,006, Rhode Island , , , , South Carolina 46 1,626, South Dakota , , , , Tennessee 95 2,458, ,434, , , , Texas 254 7,507, ,410, , , , Utah , , , , , Vermont , , , Virginia 134 3,223, ,198, , , , Washington 39 2,885, ,859, , , , West Virginia , , , , , Wisconsin 1,910 3,009,491 1,880 2,992,340 1,897 45, , ,482 1, , , Wyoming , , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 1,990, Virgin Islands 1 31,391 1 Total 6, ,862,353 6, ,968,736 6,290 1,160, , ,289 2, , , ,872 4, , ,430 Maximum 1,910 12,359,633 1,880 12,266,320 1, , , , ,522 1, , ,493 Average 119 2,299, ,325, , , , Minimum 1 31, , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 12

151 EAC Election Day Survey Over and Undervotes 2004 General Election Overvotes and Undervotes Presidential Drop-off Presidential Overvotes Presidential Undervotes Updated: 09/19/ :05:59 Total Number Percent Total Percent Percent Total Percent Percent Election Total Ballots Not Voting Not Voting Overvotes President Overvotes of Undervotes President Undervotes of Administration Ballots For For President For President For OverVotes of Total Over & For Undervotes of Total Over & Code Name Jurisdictions Counted Cases President Cases (Drop-off) (Drop-off) Cases Exception President Cases Total Ballots Cases Undervotes Cases President Cases Total Ballots Cases Undervotes Cases Election Administration Voting Equipment Used in 2004 General Election None / Unknown 908 9,922, ,782, , , , Punch card ,938, ,202, , , , Lever ,981, ,663, , , Paper 1,734 2,172,234 1,727 2,135,810 1,732 29, , , , Optical scan 2,541 49,661,061 2,524 48,141,235 2, , , ,128 1, , , ,631 2, , ,002 Electronic ,295, ,285, , , , Multiple Systems 123 8,891, ,758, , , , Changed Voting Equipment Since 2000 General Election Yes 1,753 35,479,523 1,739 34,381,658 1, , , , ,133 1, , ,163 No 4,815 86,382,830 4,749 79,587,078 4, , , ,609 1, , , ,739 3, , ,267 State Wide Voter Registration System in Place Yes 1,335 27,317,939 1,322 25,489,478 1, , , , ,597 1, , ,099 No 5,233 94,544,414 5,166 88,479,258 5, , , ,205 1, , , ,275 3, , ,331 Election Day Registration Yes 2,823 8,152,145 2,792 8,077,465 2, , , , ,612 1, , ,624 No 3, ,710,208 3, ,891,271 3,484 1,055, , ,432 2, , , ,260 2, , ,806 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 44,662,901 1,123 41,055,629 1, , , , , In Precinct Only 4,350 69,964,775 4,312 67,758,097 4, , , ,171 1, , , ,752 3, , ,308 None 1,056 7,234,677 1,053 5,155, , , , No Excuse Absentee Balloting Yes 3,781 46,531,514 3,731 46,108,172 3, , , ,934 1, , , ,041 2, , ,639 No 2,787 75,330,839 2,757 67,860,564 2, , , ,355 1, , , ,831 1, , ,791 Early Voting Allowed Yes 1,701 50,903,807 1,681 50,439,831 1, , , ,489 1, , , ,972 1, , ,412 No 4,867 70,958,546 4,807 63,528,905 4, , , , ,900 3, , ,018 Covered By Section 203, Language Minority Requirements Yes ,287, ,986, , , , No 6,100 87,574,692 6,038 79,982,072 5, , , ,210 1, , , ,186 4, , ,123 Covered By Section 5 of Voting Rights Act Yes ,429, ,638, , , , No 5,688 94,432,928 5,616 88,330,456 5, , , ,447 1, , , ,711 3, , ,804 Election Data Services, Inc. Confidential 9/22/05 Page 2 of 12

152 EAC Election Day Survey Over and Undervotes 2004 General Election Overvotes and Undervotes Presidential Drop-off Presidential Overvotes Presidential Undervotes Updated: 09/19/ :05:59 Total Number Percent Total Percent Percent Total Percent Percent Election Total Ballots Not Voting Not Voting Overvotes President Overvotes of Undervotes President Undervotes of Administration Ballots For For President For President For OverVotes of Total Over & For Undervotes of Total Over & Code Name Jurisdictions Counted Cases President Cases (Drop-off) (Drop-off) Cases Exception President Cases Total Ballots Cases Undervotes Cases President Cases Total Ballots Cases Undervotes Cases Demographics Region Northeast 1,710 20,812,375 1,687 17,662,003 1, , , , , South 1,423 42,266,877 1,417 40,376,851 1, , , ,528 1, , ,717 1, , ,043 Midwest 2,902 31,316,030 2,871 30,710,408 2, , , , ,859 2, , ,301 West ,445, ,219, , , , Territories 113 2,021, Urban to Rural Urban ,675, ,592, , , , Suburban ,263, ,993, , , , Small Towns 1,710 30,364,561 1,685 28,304,035 1, , , , ,665 1, , ,086 Rural 3,307 13,536,721 3,269 13,078,933 3, , , ,267 1, , , ,615 2, , ,181 Not Available - Territories 113 2,021, Size of Jurisdiction (VAP) < 1,000 1, ,024 1, ,085 1,754 6, , ,713 1, , ,033 >=1,000 to <3,500 1,165 1,630,543 1,162 1,597,029 1,164 40, , , >=3,500 to <10,000 1,043 4,256,986 1,038 4,156,692 1,029 60, , , , >=10,000 to <50,000 1,704 21,817,391 1,689 20,865,743 1, , , , ,906 1, , ,312 >=50,000 to <250, ,587, ,211, , , , >=250,000 to <1,000, ,485, ,329, , , , >=1,000, ,427, ,168, , , , Not Available 144 2,022, , Race and Ethnicity Predominently NH White 6, ,362,361 6, ,734,482 6,127 1,055, , , , , ,582 4, , ,330 Predominently NH Black 85 2,117, ,960, , , , Predominently NH Native American , , , , Predominently Hispanic 50 5,209, ,122, , , , Not Available 145 2,046, , Median Income < $25, ,488, ,440, , , , >=$25,000 to <$30, ,685, ,455, , , , >=$30,000 to <$35,000 1,372 14,312,622 1,356 13,179,325 1, , , , , >=$35,000 to <$40,000 1,215 26,144,458 1,204 24,868,484 1, , , , , >=$40,000 to <$45, ,227, ,994, , , , >=$45,000 to <$50, ,037, ,514, , , , >=$50,000 1,180 30,943,728 1,173 29,512,042 1, , , , , Not Available 151 2,022, , High School Education < 60% , , , , , >=60% to <70% ,083, ,719, , , , >=70% to <80% 1,646 31,406,406 1,630 29,496,147 1, , , , ,650 1, , ,125 >=80% to <90% 3,111 65,275,468 3,074 61,770,600 3, , , , ,795 2, , ,190 >=90% ,099, ,023, , , , Not Available 151 2,046, , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 12

153 EAC Election Day Survey Over and Undervotes 2004 General Election Overvotes and Undervotes Presidential Drop-off Presidential Overvotes Presidential Undervotes Updated: 09/19/ :05:59 Total Number Percent Total Percent Percent Total Percent Percent Election Total Ballots Not Voting Not Voting Overvotes President Overvotes of Undervotes President Undervotes of Administration Ballots For For President For President For OverVotes of Total Over & For Undervotes of Total Over & Code Name Jurisdictions Counted Cases President Cases (Drop-off) (Drop-off) Cases Exception President Cases Total Ballots Cases Undervotes Cases President Cases Total Ballots Cases Undervotes Cases Political Battleground States in 2004 Presidential Election Yes 3,093 43,980,255 3,028 40,593,153 2, , , , ,572 2, , ,259 No 3,475 77,882,098 3,460 73,375,583 3, , , , , , ,300 2, , ,171 Margin of Victory in 2004 Presidential Election < 2.5% ,753, ,520, , , , >=2.5% to < 5.0% 476 8,077, ,840, , , , >=5.0% to < 7.5% 510 9,931, ,790, , , , >=7.5% to < 10.0 % 429 6,126, ,948, , , , >=10.0 % 4,492 84,945,042 4,448 80,860,548 4, , , ,286 1, , , ,706 3, , ,130 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 47,293,906 3,083 44,578,904 3, , , ,421 1, , , ,770 2, , ,160 Bush 50% to 55% ,343, ,379, , , , Bush < 50% 136 1,386, ,318, , , Kerry < 50% 150 3,447, ,423, , , , Kerry 50% to 55% ,109, ,319, , , , Kerry > 55% 1,161 33,249,808 1,152 31,936,516 1, , , , , Tied 25 9, , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 12

154 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions 01 Alabama Alaska 1 04 Arizona Arkansas California Colorado Connecticut Delaware 3 11 District of Columbia 1 12 Florida Georgia Hawaii 5 16 Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin 1, Wyoming American Samoa 1 66 Guam 1 72 Puerto Rico Virgin Islands 1 Total 6,568 Maximum 1,910 Average 119 Minimum 1 Overvotes and Undervotes U.S. Senatorial Drop-off U.S. Senatorial Overvotes U.S. Senatorial Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes US Senate Overvotes of Undervotes US Senate Undervotes of For For US Senate For US Senate For OverVotes of Total Over & For Undervotes of Total Over & US Senate Cases (Drop-off) (Drop-off) Cases Exception US Senate Cases Total Ballots Cases Undervotes Cases US Senate Cases Total Ballots Cases Undervotes Cases 1,751, , , , , ,932, , , , ,040, , , , ,808, , , , ,107, , , ,424, , , ,429, , , ,222, , , , , , , , , ,998, , , , ,428, , ,462, , , , ,129, , , ,724, , ,848, , , ,323, , , ,706, , , , , , , , , ,702, , ,420, , , , , , ,427, , , , ,455, , , ,780, , , , , , , , , , , , , ,818, , , , ,869,954 1, , , , ,990, , , ,486,597 4,377 5,676, , ,100 1, , ,488, ,808,639 1,897 1,990, , , , ,242, , , , Election Data Services, Inc. Confidential 9/22/05 Page 5 of 12

155 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions Election Administration Voting Equipment Used in 2004 General Election None / Unknown 908 Punch card 260 Lever 394 Paper 1,734 Optical scan 2,541 Electronic 608 Multiple Systems 123 Changed Voting Equipment Since 2000 General Election Yes 1,753 No 4,815 State Wide Voter Registration System in Place Yes 1,335 No 5,233 Election Day Registration Yes 2,823 No 3,745 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 In Precinct Only 4,350 None 1,056 No Excuse Absentee Balloting Yes 3,781 No 2,787 Early Voting Allowed Yes 1,701 No 4,867 Covered By Section 203, Language Minority Requirements Yes 468 No 6,100 Covered By Section 5 of Voting Rights Act Yes 880 No 5,688 Overvotes and Undervotes U.S. Senatorial Drop-off U.S. Senatorial Overvotes U.S. Senatorial Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes US Senate Overvotes of Undervotes US Senate Undervotes of For For US Senate For US Senate For OverVotes of Total Over & For Undervotes of Total Over & US Senate Cases (Drop-off) (Drop-off) Cases Exception US Senate Cases Total Ballots Cases Undervotes Cases US Senate Cases Total Ballots Cases Undervotes Cases 4,837, ,277, , , ,160, , , , ,135, , , ,384,461 1,106 55, , , ,045,332 1,589 1,125, , , ,015, ,329, , , , ,594, , , ,799,451 1,434 1,031, , , , ,687,146 2,943 4,645, , ,549 1, , ,613, ,740, , , , ,746,368 3,701 5,071, , ,295 1, , ,997, ,030,935 2, , , , ,455,662 2,198 5,346, , ,977 1, , ,212, ,842,876 1,004 1,491, , , ,351, ,141,833 3,037 2,048, , , ,032, ,501, ,136, , ,268,719 3,041 1,797, , , ,567, ,217,878 1,336 3,878, , , , ,689,513 1,180 1,514, , ,512 1, , ,310, ,797,084 3,197 4,162, , , ,177, ,303, ,089, , , ,183,093 4,228 4,587, , ,465 1, , ,652, ,521, , , , ,965,170 4,016 5,106, , ,684 1, , ,953, Election Data Services, Inc. Confidential 9/22/05 Page 6 of 12

156 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions Demographics Region Northeast 1,710 South 1,423 Midwest 2,902 West 420 Territories 113 Urban to Rural Urban 567 Suburban 871 Small Towns 1,710 Rural 3,307 Not Available - Territories 113 Size of Jurisdiction (VAP) < 1,000 1,761 >=1,000 to <3,500 1,165 >=3,500 to <10,000 1,043 >=10,000 to <50,000 1,704 >=50,000 to <250, >=250,000 to <1,000, >=1,000, Not Available 144 Race and Ethnicity Predominently NH White 6,264 Predominently NH Black 85 Predominently NH Native American 24 Predominently Hispanic 50 Not Available 145 Median Income < $25, >=$25,000 to <$30, >=$30,000 to <$35,000 1,372 >=$35,000 to <$40,000 1,215 >=$40,000 to <$45, >=$45,000 to <$50, >=$50,000 1,180 Not Available 151 High School Education < 60% 126 >=60% to <70% 661 >=70% to <80% 1,646 >=80% to <90% 3,111 >=90% 873 Not Available 151 Overvotes and Undervotes U.S. Senatorial Drop-off U.S. Senatorial Overvotes U.S. Senatorial Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes US Senate Overvotes of Undervotes US Senate Undervotes of For For US Senate For US Senate For OverVotes of Total Over & For Undervotes of Total Over & US Senate Cases (Drop-off) (Drop-off) Cases Exception US Senate Cases Total Ballots Cases Undervotes Cases US Senate Cases Total Ballots Cases Undervotes Cases 9,098, , , ,215, , , , ,724,317 2, , , , , ,417, ,064, , , , ,991, ,244, ,502, , , ,314, , , , ,398,374 1, , , , , ,499,249 2, , , ,568 1, , , , ,991, , , , ,092, , , ,393, , , , ,477, , , , ,055, , , , ,242, , , , ,664, ,055, , , , ,991, ,276, ,413, ,114 1, ,224, ,337, , , , , , ,729, , , , , ,991, , , , ,199, , , , ,865, , , , ,552, , , , ,228, , , , ,040, , , , ,759, , , , , ,991, , , , ,482, , , , ,778, , , , ,838, ,035, , ,404, ,993, , , , , ,991, Election Data Services, Inc. Confidential 9/22/05 Page 7 of 12

157 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions Political Battleground States in 2004 Presidential Election Yes 3,093 No 3,475 Margin of Victory in 2004 Presidential Election < 2.5% 515 >=2.5% to < 5.0% 476 >=5.0% to < 7.5% 510 >=7.5% to < 10.0 % 429 >=10.0 % 4,492 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 Bush 50% to 55% 982 Bush < 50% 136 Kerry < 50% 150 Kerry 50% to 55% 872 Kerry > 55% 1,161 Tied 25 Overvotes and Undervotes U.S. Senatorial Drop-off U.S. Senatorial Overvotes U.S. Senatorial Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes US Senate Overvotes of Undervotes US Senate Undervotes of For For US Senate For US Senate For OverVotes of Total Over & For Undervotes of Total Over & US Senate Cases (Drop-off) (Drop-off) Cases Exception US Senate Cases Total Ballots Cases Undervotes Cases US Senate Cases Total Ballots Cases Undervotes Cases 31,048, ,164, , , ,438, ,512, ,136 1, ,549, ,307, , , , ,912, , , , ,359, , , , ,412, , , , ,455,145 3,044 2,605, , ,935 1, , ,712, ,588,738 2,176 1,215, , ,789 1, , , ,240, , , , ,114, , , ,476, , , , ,167, , , , ,860, ,256, , , , Election Data Services, Inc. Confidential 9/22/05 Page 8 of 12

158 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions 01 Alabama Alaska 1 04 Arizona Arkansas California Colorado Connecticut Delaware 3 11 District of Columbia 1 12 Florida Georgia Hawaii 5 16 Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin 1, Wyoming American Samoa 1 66 Guam 1 72 Puerto Rico Virgin Islands 1 Total 6,568 Maximum 1,910 Average 119 Minimum 1 U.S. Congressional Drop-off Overvotes and Undervotes U.S. Congressional Overvotes Cases = Number of Jurisdictions Reporting Subject Matter U.S. Congressional Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes Congress Overvotes of Undervotes Congress Undervotes of For For Congress For Congress For OverVotes of Total Over & For Undervotes of Total Over & Congress Cases (Drop-off) (Drop-off) Cases Exception Congress Cases Total Ballots Cases Undervotes Cases Congress Cases Total Ballots Cases Undervotes Cases 1,638, , , , , ,869, , , , , , , ,265, ,155, , , ,040, , , ,428, , , , , , , , , ,627, ,887, , ,256, ,060, , , , , , , , ,866, , ,431, , , ,156, , , ,635, , ,035, , , , , ,228, , , ,472, , , ,628, , , ,721, , , ,116, , ,749, , , , , , , , , , , , , , ,284, , , , , , , ,819, , ,409, , , , , , ,418, , , ,772, , , , , , , , , , , ,836, , , , , , , , , , , ,548, , , ,729, , , , , , , ,815, , , , , , , ,959, , , , , , ,338, ,669, , ,077, ,486 10,265, ,155, , , ,818 1,798, , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 9 of 12

159 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions Election Administration Voting Equipment Used in 2004 General Election None / Unknown 908 Punch card 260 Lever 394 Paper 1,734 Optical scan 2,541 Electronic 608 Multiple Systems 123 Changed Voting Equipment Since 2000 General Election Yes 1,753 No 4,815 State Wide Voter Registration System in Place Yes 1,335 No 5,233 Election Day Registration Yes 2,823 No 3,745 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 In Precinct Only 4,350 None 1,056 No Excuse Absentee Balloting Yes 3,781 No 2,787 Early Voting Allowed Yes 1,701 No 4,867 Covered By Section 203, Language Minority Requirements Yes 468 No 6,100 Covered By Section 5 of Voting Rights Act Yes 880 No 5,688 U.S. Congressional Drop-off Overvotes and Undervotes U.S. Congressional Overvotes Cases = Number of Jurisdictions Reporting Subject Matter U.S. Congressional Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes Congress Overvotes of Undervotes Congress Undervotes of For For Congress For Congress For OverVotes of Total Over & For Undervotes of Total Over & Congress Cases (Drop-off) (Drop-off) Cases Exception Congress Cases Total Ballots Cases Undervotes Cases Congress Cases Total Ballots Cases Undervotes Cases President, U.S. Senate, and U.S. House of Representatives Drop-Off excluded where Drop-Off = 0. 5,193, , , , ,618, , , , ,439, ,334, , ,464, , , ,376, ,227, , ,296, ,862 19,563, ,402, , ,542, ,682, ,008, , ,546, ,195, , ,001, ,253 59,791, ,474, , ,075, ,233 22,199, ,872, , ,461, ,045 64,138, ,797, , ,615, ,441 7,711, , , ,941 78,626, ,201, , ,722, ,545 30,907, ,582, , ,991, ,472, ,808, , ,939, ,389 6,957, , , , ,200, ,663, , ,504, ,909 46,137, ,006, , ,572, ,577 41,451, ,674, , ,873, ,355 44,886, ,994, , ,203, ,131 25,887, ,966, , ,982, ,450, ,702, , ,094, ,207 21,772, ,473, , ,676, ,565, ,196, , ,400, ,885 Election Data Services, Inc. Confidential 9/22/05 Page 10 of 12

160 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions Demographics Region Northeast 1,710 South 1,423 Midwest 2,902 West 420 Territories 113 Urban to Rural Urban 567 Suburban 871 Small Towns 1,710 Rural 3,307 Not Available - Territories 113 Size of Jurisdiction (VAP) Overvotes and Undervotes < 1,000 1, , , , ,336 >=1,000 to <3,500 1,165 1,502, , , >=3,500 to <10,000 1,043 3,742, , , , >=10,000 to <50,000 1,704 16,947, ,345, , ,014, ,094 >=50,000 to <250, ,517, ,077, , ,167, >=250,000 to <1,000, ,788, ,043, , ,509, >=1,000, ,270, ,655, , ,013, Not Available 144 1,992, , , , Race and Ethnicity Predominently NH White 6,264 Predominently NH Black 85 Predominently NH Native American 24 Predominently Hispanic 50 Not Available 145 Median Income < $25, ,191, , , >=$25,000 to <$30, ,234, , , >=$30,000 to <$35,000 1,372 11,170, , , , >=$35,000 to <$40,000 1,215 15,829, ,768, , , >=$40,000 to <$45, ,709, ,574, , ,290, >=$45,000 to <$50, ,594, ,770, , , >=$50,000 1,180 22,615, ,294, , ,397, Not Available 151 1,992, , , , High School Education U.S. Congressional Drop-off U.S. Congressional Overvotes Cases = Number of Jurisdictions Reporting Subject Matter U.S. Congressional Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes Congress Overvotes of Undervotes Congress Undervotes of For For Congress For Congress For OverVotes of Total Over & For Undervotes of Total Over & Congress Cases (Drop-off) (Drop-off) Cases Exception Congress Cases Total Ballots Cases Undervotes Cases Congress Cases Total Ballots Cases Undervotes Cases 11,995, ,086, , , ,841, ,367, , ,303, ,836, ,350, , , ,309 22,676, ,832, , ,327, ,989, , , , ,198, ,123, , ,061, ,915, ,118, , ,337, ,603, ,682, , ,140, ,631, , , , ,206 1,989, , , , ,272, ,597, , ,514, ,281 1,377, , , , , , ,549, , , , ,016, , , , < 60% , , , >=60% to <70% 661 8,292, , , , >=70% to <80% 1,646 20,677, ,034, , ,271, ,055 >=80% to <90% 3,111 44,262, ,515, , ,480, ,198 >=90% 87310,311, , , , Not Available 151 2,016, , , , Election Data Services, Inc. Confidential 9/22/05 Page 11 of 12

161 EAC Election Day Survey Over and Undervotes 2004 Genera Updated: 09/19/ :05:59 Election Administration Code Name Jurisdictions Political Battleground States in 2004 Presidential Election Overvotes and Undervotes Yes 3,093 30,660, ,396, , ,515, ,335 No 3,475 55,677, ,273, , ,562, ,151 Margin of Victory in 2004 Presidential Election < 2.5% 515 >=2.5% to < 5.0% 476 >=5.0% to < 7.5% 510 >=7.5% to < 10.0 % 429 >=10.0 % 4,492 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 Bush 50% to 55% 982 Bush < 50% 136 Kerry < 50% 150 Kerry 50% to 55% 872 Kerry > 55% 1,161 Tied 25 U.S. Congressional Drop-off U.S. Congressional Overvotes Cases = Number of Jurisdictions Reporting Subject Matter U.S. Congressional Undervotes Total Number Percent Total Percent Percent Total Percent Percent Ballots Not Voting Not Voting Overvotes Congress Overvotes of Undervotes Congress Undervotes of For For Congress For Congress For OverVotes of Total Over & For Undervotes of Total Over & Congress Cases (Drop-off) (Drop-off) Cases Exception Congress Cases Total Ballots Cases Undervotes Cases Congress Cases Total Ballots Cases Undervotes Cases 7,345, ,257, , , ,877, , , , ,979, ,145, , , ,322, , , ,816, ,531, , ,806, ,050 36,235, ,469, , ,015, ,035 12,785, ,067, , , ,008, , , ,418, , , , ,476, ,887, , , ,413, ,722, , ,598, , Election Data Services, Inc. Confidential 9/22/05 Page 12 of 12

162 Chapter 9 Voting Equipment Usage Table 9 presents data from the Election Day Survey on the type of voting equipment used by election jurisdictions during the November 2, 2004, election. The survey asked for a listing of the type and manufacturer of voting systems in use; the number of units for each system; software versions, if applicable; and whether the voting systems previously had been used in a federal election. Applicability and Coverage State responses to the five parts of question 13 on voting equipment usage were mixed. Some states responded to each part, while others provided a single response that had to be split apart and reclassified. Some states provided important information that was not requested e.g, name of voting equipment hardware or ballot tabulation method (e.g., precinct or central-count). Nine states did not respond to question 13 at all. All information that states provided on voting equipment, including information not requested by the survey, was standardized following a format established by National Association of State Election Directors (NASED) for qualifying voting systems. 1 Therefore, the tabulated version of question 13 has nine parts: 1) Company 6) Voting equipment type 2) Voting system 7) Ballot tabulation method 3) Software 8) Previous use at a federal election 4) Voting equipment hardware 9) Number of units 5) Voting equipment hardware version Only data on voting equipment type is generally complete. For states that did not respond to question 13, information on equipment type was obtained from media reports or voting equipment summaries published by the National Association of Secretaries of State on the Web. In other instances, voting equipment type was determined from other information on the survey for voting equipment hardware. About 30 states identified voting equipment manufacturers, and about 20 states provided information on the number of units and previous use at a federal election. The survey results on the different types of voting equipment used by local election jurisdictions are presented in chapter 9. Also included in this chapter is information on voting equipment hardware and manufacturers. The results on the number of units and ballot tabulation methods are in chapter 10. The results on voting equipment malfunctions are in chapter 11. There are five generic types of voting equipment: (1) punch card, (2) optical scan, (3) electronic, (4) lever machine, and (5) paper ballot. A sixth category, mixed, is for jurisdictions using more than 1 Under the Help America Vote Act of 2002, the U.S. Election Assistance Commission (EAC) is responsible for the certification of voting systems. That function was previously carried out by NASED.

163 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-2 September 27, 2005 one type of voting equipment. The punch card category includes DataVote and Votomatic systems. 2 Electronic refers to direct recording electronic (DRE) devices utilizing touch screens, push buttons, or select wheels for voters to indicate their preferences. The following is a comprehensive listing of voting equipment hardware that local election jurisdictions reported were used at the November 2, 2004, general election. The listing is by company. In some instances, survey respondents reported the names of voting equipment distributors instead of manufacturers. Some of the companies Automatic Voting Machine Corp. (AVM), for example are no longer in business. Because of licensing or other agreements, some voting equipment hardware Optech Eagle, for example was manufactured or sold by two or more companies. Table 9a. Voting Equipment for the November 2004 General Election Company Hardware Type Airmac Technology Systems, Inc. MR 210 Optical scan Automatic Voting Machine Corp. (AVM) AVM Lever Predecessor of Sequoia Voting Systems AVM-NP Lever AVM-NS Lever AVM-POM 40 Lever AVM-POM 50 Lever AVM-Printomatic Lever AVM-RS Lever Advanced Voting Solutions (AVS) AVM-NS Lever Formerly, Shoup Voting Solutions, Inc. Pre- AVM-RS Lever decessor companies include R.F. Shoup Corp. WINscan Optical scan WINvote Electronic Computing Devices Canada Elex Voting System Electronic Danaher Controls ELECTronic 1242 Electronic Division of Danaher Corp. Shouptronic 1242 Electronic DFM Associates Mark-A-Vote Optical scan Diebold Election Systems AccuTouch Electronic Predecessor companies include Global Election AccuTouch (Global) Electronic Systems Inc. and Data Information Management AccuVote ES-2000 Optical scan Systems (DIMS) AccuVote ES-2000 (Global) Optical scan AccuVote-OS Optical scan AccuVote-OS (Global) Optical scan AccuVote-TS Electronic AccuVote-TS R6 Electronic Global NCS 5 Optical scan Table 9a. Voting Equipment 2004 (cont.) 2 A Votomatic ballot is prescored and printed only with numbered voting positions. A Votomatic ballot is inserted into a frame to which an attached booklet identifies candidates or answers to ballot questions for each voting position. A stylus is used to punch out prescored chads at voting positions. A Datavote ballot card is printed with a candidate name or answer to a ballot question at each voting position. A Datavote ballot card is inserted into a frame with a movable punching device. Voters align the punching device with a candidate name or answer to a ballot question, and then press the device to punch out a hole in the card at the appropriate voting position. Because candidate names are printed on the actual Datavote card, a usual election involves multiple cards that a voter must cast to complete their ballot.

164 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-3 September 27, 2005 Company Hardware Type Data Information Management Systems DataVote Punch Card Predecessor of Diebold Inc. ES&S (Election Systems & Software) BRC P-III Punch card Predecessor companies include American Cardamation (CES) Punch card Information Systems, Inc. (AIS), Business CES Votomatic Punch card Records Corp. (BRC), and Computer Election Dell Ultra Scan Punch card Systems (CES) DOC 600 Punch card ETP-IV (BRC) (unknown) ivotronic Electronic Model 100 Optical scan Model 115 Optical scan Model 115 (AIS) Optical scan Model 150 Optical scan Model 150 (AIS) Optical scan Model 315 Optical scan Model 315 (AIS) Optical scan Model 550 Optical scan Model 550 (AIS) Optical scan Model 650 Optical scan Optech Optical scan Optech 1 (CES) Optical scan Optech Eagle Optical scan Optech II Optical scan Optech II (BRC) Optical scan Optech III Optical scan Optech III (BRC) Optical scan Optech III-P Optical Scan Optech III-P (BRC) Optical scan Optech III-P Eagle Optical scan Optech IV-C Optical scan Optech IV-C (BRC) Optical scan Optech IV-C 200 Optical scan Optech IV-C 200 (BRC) Optical scan Optech IV-C 400 Optical scan Optech IV-C 400 (BRC) Optical scan PBC 2100 Punch card PBC 5 Punch card PBC 6 Punch card PBC III-D Punch card PBC III-D (BRC) Punch card PC-BT Punch card Pollstar Votronic Punch card V-200 Electronic Votomatic Punch card Votomatic (CES) Punch card Table 9a. Voting Equipment 2004 (cont.) Company Hardware Type ES&S (cont.) Votomatic III Punch card

165 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-4 September 27, 2005 Votronic Votronic I Votronic II Electronic Electronic Electronic Fidlar AccuVote ES-2000 Optical scan Predecessor companies include Fidlar Double- AccuVote-OS Optical scan Day Inc., Fidlar & Chambers Co. and Govern- DIS Punch card mental Business Systems Inc. (GBS) EV 2000 Electronic Governmental Data Systems (GDS) ATS MR 200 Optical scan Hart InterCivic Inc. Ballot Now Optical scan eslate Electronic International Election Systems (IES) Shoup Model 2.5 Lever Shoup Model 3.2 Lever MicroVote General Corp. Infinity Electronic MEMS (unknown) MV-464 Electronic Peripheral Dynamics Inc. (PDI) PDI 6111 HT Punch card R.F. Shoup Corp. Shoup Lever Predecessor of Shoup Voting Solutions/AVS Shouptronic 1242 Electronic Sequoia Voting Systems AVC Advantage Electronic Predecessor companies include Sequoia AVC Edge Electronic Pacific Voting Systems, Inc. and Automatic AVM Lever Voting Machine Corp. DataVote Punch Card Optech Eagle Optical scan Optech III-P Optical scan Optech III-P Eagle Optical scan Optech Insight Optical scan Optech IV-C Optical scan AVM Printomatic-30 Lever Triad Governmental Systems Inc. ELECtab Punch Card Unilect Corp. Patriot Electronic Votec Corp. Votomatic III-A Punch card Votomatic III-P Punch card Voting Technologies International (VTI) VoT Ware Electronic Webb Systems BCCS 228 Punch card BCCS 312 Punch card (company unknown) Benton BC-1000 Punch card Documation Punch card Documentation M-1000 Punch card Documentation M-200 Punch card Documentation M-600L Punch card ETNet Punch card Only about half of the survey respondents provided the names of the manufacturers of voting systems used at the November 2004 election. The following table identifies the manufacturers and, in some instances, the distributors, of voting systems used by local election jurisdictions in 2004 and is sorted by the number of jurisdictions covered.

166 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-5 September 27, 2005 Table 9b. Voting Equipment Manufacturers, November 2004 General Election Voting Equipment Manufacturer Jurisdictions ES&S (Election Systems & Software) 1,638 Diebold Election Systems 1,444 Danaher Controls 236 Sequoia Voting Systems (including AVM) 201 Fidlar 180 Microvote General Corp. 169 Advanced Voting Solutions (incl. R.F.Shoup) 44 IES (International Election Systems) 33 Hart InterCivic Inc. 25 Unilect Corp. 25 Webb Systems 9 DFM Associates 8 Triad Governmental Systems Inc. (Triad GSI) 7 Votec Corp. 3 AirMac Technologies Inc. 1 Computing Devices Canada (CDC) 1 Governmental Data Systems (GDS) 1 Peripheral Dynamics Inc. (PDI) 1 Voting Technologies International (VTI) 1 (Manufacturer not indicated) 3,970 Historical Context Prior to 1980, no one kept track of what kind of voting equipment was used in all jurisdictions in the United States. While state laws may provide some parameters for the type of voting equipment that is allowed in a state, the final decision has traditionally been left to local election administrators. In 1979, the Federal Election Commission s Office of Election Administration (the predecessor to the new U.S. Election Assistance Commission [EAC]) contracted with Election Data Services, Inc. to conduct regional workshops on a variety of election administration topics, including voting equipment. As part of those workshops, Election Data Services began compiling information on the type of voting equipment that was used in all jurisdictions. In what became a labor of love (or insanity), we continued to keep track of voting equipment usage around the nation. With each subsequent general election, we saved an archive of the database, which allowed us to monitor change over time. Figures 9.1 and 9.2 show the percentage of counties and percentage of registered voters that have used each voting equipment type from 1980 to the present time. A comparison of information on the two charts at any point in time provides important insights on election administration in this country. For example, while nearly 10 percent of counties in 2004 used paper ballots, they amounted to less than one percent of the registered voters. This indicates that very small counties are the predominant users of paper ballots. The opposite percentage comparison can be seen in punch card and electronic voting machine usage, an indication that these systems are used by the larger jurisdictions in the nation. Figure 9.1 Voting Equipment , Percent of Counties

167 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-6 September 27, 2005 Figure 9.2 Voting Equipment , Percent of Registered Voters

168 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-7 September 27, 2005 For much of this country s history, voters have used paper ballots. As the country grew and became more urbanized, the task of counting paper ballots took longer. With the Industrial Revolution, a mechanical way was found to produce instantaneous election results the lever machine. Mechanical lever machines were invented in the 1890s, and their use in elections grew rapidly over the next 70 years. Lever machines combined the casting, recording, and counting of votes in one apparatus. And it is interesting to note in light of the current controversy over electronic voting that for all those 70 years, voters were not receiving, nor were election officials counting, physical ballots. Precincts then tended to be smaller in size because the high cost of lever machines prevented election officials from placing large numbers of these devices in each precinct. By the middle of the 20th century, the main source of polling place judges housewives had begun moving into the workforce. As a result of this loss in manpower (or womanpower), election officials looked to cutting the overall number of precincts and increasing the size of the remaining polling places. Punch card voting systems, first used in 1964, were a popular solution to this problem. Suburban and urban communities around the nation soon found that the cost of 10 punch card devices was similar to just two lever machines, allowing election officials to create larger precincts. While bigger counties began to adopt punch cards, smaller counties needed a solution that would allow them to continue to use paper ballots, but tally election results more quickly. This led to the development of optical scan devices for voting in the 1970s and 1980s. With the advent of computers and the need to replace aging mechanical lever machines, the 1970s also saw the introduction of electronic voting systems. Early electronic voting devices looked much like lever machines, with push buttons replacing levers on a large panel. Newer DREs, resembling ATMs (automatic teller machines), had touch-screen panels and keypads for entering write-in votes. Voter preferences went directly into electronic storage, usually without a paper record of the voter s intent. Survey Results Table 9 presents data on voting equipment usage from question 13 on the Election Day Survey. In the table, both the number of registered voters and the count of jurisdictions, using each of six different types of voting equipment (including multiple systems), is calculated. Because the EAC dataset is incomplete, a seventh category of Unknown is also shown. Because the information is so wide, the table spans multiple pages. The column headings in Table 9 are as follows:

169 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-8 September 27, 2005 Table 9 Column Headings. Voting Equipment Usage Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question Total Registration 5 Number of Juris. Using Punch card Equipment 6 % of Juris. Using Punch card Equipment 7 Regis. in Juris. Using Punch card Equipment Number of active and inactive registered voters, number of persons who voted on Election Day in six states, and VAP data for North Dakota and jurisdictions in Wisconsin that do not have voter registration, from col. 4 of table 2 Number of jurisdictions that responded to survey question 13 and reported the use of punch cards Number of jurisdictions using punch cards (col. 5) divided by the total number of election jurisdictions (col. 3) Number of registered voters in jurisdictions that reported the use of punch cards 8 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data; and jurisdictions that responded to question 13 and reported the use of punch cards 9 % of Regis. Using Punch card Equipment 10 Number of Juris. Using Optical Scan Voting Equipment 11 % of Juris. Using Optical Scan Voting Equipment 12 Regis. in Juris. Using Optical Scan Voting Equipment Number of registered voters in jurisdictions that reported the use of punch cards (col. 7) divided by the total number of registered voters (col. 3) Number of jurisdictions that responded to survey question 13 and reported the use of optical scan equipment Number of jurisdictions using optical scan equipment (col. 10) divided by the total number of election jurisdictions (col. 3) Number of registered voters in jurisdictions that reported the use of optical scan equipment 13 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data; and jurisdictions that responded to question 13 and reported the use of optical scan equipment 14 % of Regis. Using Optical Scan Voting Equipment Number of registered voters in jurisdictions that reported the use of optical scan equipment (col. 12) divided by the total number of registered voters (col. 3)

170 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-9 September 27, 2005 Table 9 Column Headings (cont.) Col. Heading Description 15 Number of Juris. Using Electronic Voting Equipment 16 % of Juris. Using Electronic Voting Equipment 17 Regis. in Juris. Using Electronic Voting Equipment Number of jurisdictions that responded to survey question 13 and reported the use of electronic equipment Number of jurisdictions using electronic equipment (col. 15) divided by the total number of election jurisdictions (col. 3) Number of registered voters in jurisdictions that reported the use of electronic equipment 18 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data; and jurisdictions that responded to question 13 and reported the use of electronic equipment 19 % of Regis. Using Electronic Voting Equipment 20 Number of Juris. Using Lever Machine Voting Equipment 21 % of Juris. Using Lever Machine Voting Equipment 22 Regis. in Juris. Using Lever Machine Voting Equipment Number of registered voters in jurisdictions that reported the use of electronic equipment (col. 17) divided by the total number of registered voters (col. 3) Number of jurisdictions that responded to survey question 13 and reported the use of lever machines Number of jurisdictions using lever machines (col. 20) divided by the total number of election jurisdictions (col. 3) Number of registered voters in jurisdictions that reported the use of lever machines 23 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data; and jurisdictions that responded to question 13 and reported the use of lever machines 24 % of Regis. Using Lever Machine Voting Equipment 25 Number of Juris. Using Paper Ballots Voting Equipment 26 () Extra line of space in this cell at top % of Juris. Using Paper Ballots Voting Equipment 27 Regis. in Juris. Using Paper Ballots Voting Equipment Number of registered voters in jurisdictions that reported the use of lever machines (col. 24) divided by the total number of registered voters (col. 3) Number of jurisdictions that responded to survey question 13 and reported the use of paper ballots Number of jurisdictions using paper ballots (col. 25) divided by the total number of election jurisdictions (col. 3) Number of registered voters in jurisdictions that reported the use of paper ballots

171 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-10 September 27, 2005 Table 9 Column Headings (cont.) Col. Heading Description 28 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data; and jurisdictions that responded to question 13 and reported the use of paper ballots 29 % of Regis. Using Paper Ballots Voting Equipment 30 Number of Juris. Using Mixed Voting Equipment 31 % of Juris. Using Mixed Voting Equipment 32 Regis. in Juris. Using Mixed Voting Equipment Number of registered voters in jurisdictions that reported the use of paper ballots (col. 27) divided by the total number of registered voters (col. 3) Number of jurisdictions that responded to survey question 13 and reported the use of two or more types of voting equipment Number of jurisdictions using two or more types of voting equipment (col. 30) divided by the total number of election jurisdictions (col. 3) Number of registered voters in jurisdictions that reported the use of two or more types of voting equipment 33 Cases Number of jurisdictions that responded to survey question 1, that provided Election Day registration data, or for which VAP data was substituted for voter registration data; and jurisdictions that responded to question 13 and reported the use of two or more types of voting equipment 34 % of Regis. Using Mixed Voting Equipment 35 Number of Juris., Unknown (Not Reported) 36 % of Juris., Unknown (Not Reported) 37 Regis. in Juris., Unknown (Not Reported) Number of registered voters in jurisdictions that reported the use of two or more types of voting equipment (col. 32) divided by the total number of registered voters (col. 3) Number of jurisdictions that responded to parts of survey question 13, but not the type of voting equipment used Number of jurisdictions that responded to parts of survey question 13, but not the type of voting equipment used (col. 35) divided by the total number of election jurisdictions (col. 3) Number of registered voters in jurisdictions that responded to parts of survey question 13, but not the type of voting equipment used 38 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data; and jurisdictions that responded to parts of survey question 13, but not the type of voting equipment used 39 % of Regis., Unknown (Not Reported) Number of registered voters in jurisdictions that parts of survey question 13, but not the type of voting equipment used (col. 37) divided by the total number of registered voters (col. 3)

172 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-11 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 3 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 9 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary The EAC dataset shows the following nationwide characteristics of voting equipment usage in the 2004 general election. (See Table 9c below.) As noted previously, the information is not complete for nearly 14 percent of the jurisdictions in the nation, which covers more than four percent of the registered voters. Type of Voting Equipment Table 9c. Voting Equipment Usage, 2004 General Election Number of Jurisdictions Percent of Jurisdictions Registration in Jurisdictions Percent of Registration Cases Punch Card ,691, Optical Scan 2, ,544,126 2, Electronic ,571, Lever ,662, Paper Ballots 1, ,068,977 1, Mixed Systems ,367, Unknown ,726, TOTAL 6, ,632,295 6, The data in Table 9a is illustrated in the following charts:

173 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-12 September 27, 2005 Figure 9.3. Voting Equipment Usage, 2004: Jurisdictions Number of Jurisdictions % % % 2, % 1, % % % Punch Card Optical Scan Electronic Lever Paper Ballots Mixed Systems Unknown Figure 9.4. Voting Equipment Usage, 2004: Registration Registration in Jurisdictions 13,367, % 3,068, % 21,662, % 7,726, % 15,691, % 69,544, % 43,571, % Punch Card Optical Scan Electronic Lever Paper Ballots Mixed Systems Unknown Besides being incomplete, one of the most significant differences between the data published by Election Data Services and the EAC is the geographic structure used. While Election Data Services keeps its database at the jurisdiction level, for the purposes of mapping and publishing Election Data Services has traditionally shown data for counties. This has meant that for the New England states,

174 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-13 September 27, 2005 along with Michigan, Minnesota, and Wisconsin, there is a greater existence of mixed systems in the Election Data Services results, along with a corresponding decrease in the individual voting system types. This is because the jurisdiction data is summed to the county level. In these nine states voting-equipment purchase decisions have been traditionally made by individual townships. Election Data Services has independently verified that for jurisdictions in the nation over one-quarter still use paper ballots. States While local jurisdictions have traditionally made the decision about what kind of voting equipment to purchase, more states have begun imposing a mandate that uniform equipment types be used. This was not the result of the 2000 Bush v. Gore court decision by the U.S. Supreme Court, but began years earlier in several states as a cost-savings strategic decision. The 2004 elections found 12 states and territories that had adopted uniform voting-equipment usage. The states and the voting equipment type used are: Optical scan Alaska, Arizona, Oklahoma, Oregon, and Rhode Island; electronic Delaware, Georgia, Maryland, and Nevada; lever machines New York; and mixed systems District of Columbia and Hawaii. Regions The Midwest has the largest number of jurisdictions using punch cards, but punch cards account for just 18 percent of the registered voters in the region. A plurality of jurisdictions in the Midwest use optical scan voting systems. However, the greater percentage of registered voters using optical scan systems is in the West, where over 62 percent of the registered voters mark their ballots with a pencil or pen. Because of the state of Georgia s recent adoption of electronic voting, the South has the largest usage of electronic voting systems in the country. A plurality of just over 41 percent of the registered voters in the South use electronic voting equipment. Lever machines still dominate the Northeast where over 55.4 percent of the registered voters in the region use lever devices, which were manufactured in the Northeast. Paper ballots are mainly split between the Midwest and the Northeast, but in either region they are used by less than 3.5 percent of the registered voters. Urban to Rural Over 58 percent of the urban jurisdictions in this nation use optical scan voting systems, as do a plurality of the suburban and small town communities. A plurality of the rural jurisdictions in the nation use paper ballots, closely followed by optical scan systems. Due to the distribution of registered voters in rural communities, however, over 60 percent of rural registered voters have their ballots counted by optical scan systems. Size of Jurisdiction The EAC data shows that the larger the size of a jurisdiction the more likely they are to be using electronic voting equipment. This has traditionally not been the case, as punch cards were almost exclusively used by large jurisdictions in this nation. Paper ballots are mainly concentrated in the smaller jurisdictions, with over 61 percent of the 1,761 jurisdictions that have under 1,000 voting age population (VAP) using pencil and pens, and presumably counting the ballots by hand. As jurisdictions grow in size they move to optical scan systems to ease the counting process. A clear majority of the jurisdictions that have between 3,500 and 50,000 VAP use optical scan devices.

175 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-14 September 27, 2005 Race and Ethnicity A plurality of the predominantly non-hispanic White jurisdictions in this nation use optical scan technology, closely followed by electronic systems. On the other hand, nearly two-thirds of African American voters are casting their votes on electronic systems. Over two-thirds of voters in predominantly Native American and Hispanic communities use optical scan devices. Median Income There appears to be a small relationship between income levels and the use of optical scan systems. The lower the income levels in a jurisdiction, the higher likelihood they will be using optical scan systems. Electronic machines are used to a greater degree in both the poorest and the wealthiest jurisdictions in the nation. High School Education There does not appear to be a relationship between education levels and type of voting equipment used. Section 203 Language Minority Requirements A majority of the jurisdictions covered by the Section 203 language minority requirements of the Voting Rights Act use optical scan voting systems. Due to the smaller size of these jurisdictions, however, they represent just a plurality of the registered voters. Section 5 Preclearance of Voting Procedures A plurality of the jurisdictions and registered voters covered by Section 5 of the Voting Rights Act used optical scan devices in 2004, closely followed by electronic voting equipment. Type of Voting Equipment Not applicable. Changed Voting Equipment since 2000 A plurality of the jurisdictions that changed voting equipment since 2000 went to optical scan systems. Statewide Voter Registration Database No real patterns are discernible in this subcategory. Election Day Registration A majority of registered voters in jurisdictions that allow Election Day registration are much more likely to use optical scan voting equipment. Provisional Ballot Acceptance No real patterns are discernible in this subcategory. No Excuse Absentee Balloting No real patterns are discernible in this subcategory.

176 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Usage, Page 9-15 September 27, 2005 Early Voting While a slight majority of registered voters who can utilize early voting procedures are using optical scan devices, there is a larger-than-average number that are also on electronic voting systems. Electronic voting systems are thought to be easier to manage in an early voting environment because they can maintain many different ballot styles in a jurisdiction. This would eliminate the need to keep lots of different paper ballots at each early-voting site. Battleground States No real patterns are discernible in this subcategory, although electronic voting systems were more likely to be found in nonbattleground states. Presidential Margin of Victory No real patterns are discernible in this subcategory. Red versus Blue Jurisdictions No real patterns are discernible in this subcategory, although lever machines were more likely to be used in jurisdictions carried by Kerry than by Bush. On the other hand, jurisdictions using punch cards were much more likely to support Bush than Kerry.

177 EAC Election Day Survey Voting Equipment Usage 2004 General Election Updated: 09/19/ :06:28 Voting Equipment Usage Punchcard Voting Equipment (Inc.Datavote) Optical Scan Voting Equipment Electronic Voting Equipment Election 2004 Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Administration Total of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Code Name Jurisdictions Registration Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using 01 Alabama 67 2,597, ,221, , Alaska 1 472, , Arizona 15 2,642, ,642, Arkansas 75 1,699, , , , California 58 16,646, , ,980, ,118, Colorado 64 3,101, , ,504, , Connecticut 169 1,831, Delaware 3 553, , District of Columbia 1 383, Florida 67 10,300, ,630, ,576, Georgia 159 4,248, ,248, Hawaii 5 647, Idaho , , , Illinois 110 7,195, ,914, ,281, Indiana 92 4,296, , ,580, ,090, Iowa 99 2,226, ,978, , Kansas 105 1,695, ,007, , Kentucky 120 2,794, , ,268, Louisiana 64 2,932, ,602, Maine 517 1,026, , Maryland 24 3,105, ,105, Massachusetts 351 4,098, ,871, Michigan 83 7,164, , ,273, , Minnesota 87 2,977, ,401, Mississippi 82 1,469, , ,004, , Missouri 116 4,194, ,602, ,779, Montana , , Nebraska 93 1,160, ,047, Nevada 17 1,073, ,073, New Hampshire , , New Jersey 21 5,011, , ,645, New Mexico , New York 58 11,837, North Carolina 100 5,526, , ,388, ,364, North Dakota , , Ohio 88 7,965, ,717, , Oklahoma 77 2,143, ,143, Oregon 36 2,141, ,141, Pennsylvania 67 8,366, , , ,271, Rhode Island , , South Carolina 46 2,318, , , ,001, South Dakota , , Tennessee 95 3,748, , , ,726, Texas ,098, , ,051, , Utah 29 1,278, ,257, , Vermont , , Virginia 134 4,515, , , , Washington 39 3,508, , ,277, , West Virginia 55 1,168, , , , Wisconsin 1,910 4,179, ,750, Wyoming , , , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 2,440, Virgin Islands 1 50,731 Total 6, ,265, ,767, , ,198,628 2, ,068, Maximum 1,910 16,646, ,914, ,980, ,576, Average 119 3,344, , ,729, ,484, Minimum 1 50, , , , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 8

178 EAC Election Day Survey Voting Equipment Usage 2004 General Election Updated: 09/19/ :06:28 Voting Equipment Usage Punchcard Voting Equipment (Inc.Datavote) Optical Scan Voting Equipment Electronic Voting Equipment Election 2004 Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Administration Total of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Code Name Jurisdictions Registration Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,484,493 Punch card ,767, ,767, Lever ,662,619 Paper 1,734 3,085,167 Optical scan 2,541 69,198,628 2, ,198,628 2, Electronic ,068, ,068, Multiple Systems ,997,891 Changed Voting Equipment Since 2000 General Election Yes 1,753 51,149, , ,735, ,326, No 4, ,115, ,739, , ,463,280 1, ,741, State Wide Voter Registration System in Place Yes 1,335 37,384, , ,487, ,802, No 5, ,880, ,039, , ,711,105 1, ,265, Election Day Registration Yes 2,823 10,323, , ,960, , No 3, ,941, ,180, , ,237,905 1, ,062, Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,077, ,444, ,469, ,674, In Precinct Only 4, ,336, ,771, , ,218,106 1, ,394, None 1,056 8,850, , ,511, No Excuse Absentee Balloting Yes 3,781 64,333, ,846, , ,815,566 1, ,899, No 2, ,931, ,921, , ,383,062 1, ,169, Early Voting Allowed Yes 1,701 73,710, ,690, ,307, ,493, No 4, ,554, ,076, , ,891,400 1, ,575, Covered By Section 203, Language Minority Requirements Yes ,756, ,613, ,390, ,116, No 6, ,508, ,154, , ,807,981 2, ,952, Covered By Section 5 of Voting Rights Act Yes ,868, ,874, ,254, ,617, No 5, ,396, ,893, , ,944,544 2, ,450, Election Data Services, Inc. Confidential 9/22/05 Page 2 of 8

179 EAC Election Day Survey Voting Equipment Usage 2004 General Election Updated: 09/19/ :06:28 Voting Equipment Usage Punchcard Voting Equipment (Inc.Datavote) Optical Scan Voting Equipment Electronic Voting Equipment Election 2004 Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Administration Total of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Code Name Jurisdictions Registration Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using Demographics Region Northeast 1,710 34,273, , ,208, ,917, South 1,423 62,606, ,273, ,839, ,647, Midwest 2,902 44,048, ,088, , ,046,260 1, ,987, West ,845, ,415, ,104, ,516, Territories 113 2,490,862 Urban to Rural Urban ,441, ,195, ,979, ,868, Suburban ,552, ,975, ,938, ,784, Small Towns 1,710 44,193, ,824, ,438, ,571, Rural 3,307 19,586, ,771, , ,841,737 1, ,843, Not Available - Territories 113 2,490,862 Size of Jurisdiction (VAP) < 1,000 1, , , >=1,000 to <3,500 1,165 2,182, , , , >=3,500 to <10,000 1,043 5,966, , ,906, , >=10,000 to <50,000 1,704 31,472, ,768, ,102, ,839, >=50,000 to <250, ,992, ,973, ,063, ,865, >=250,000 to <1,000, ,396, ,995, ,260, ,083, >=1,000, ,867, ,794, ,752, ,704, Not Available 144 2,492,279 Race and Ethnicity Predominently NH White 6, ,662, ,677, , ,432,043 2, ,073, Predominently NH Black 85 3,098, , , ,935, Predominently NH Native American , , Predominently Hispanic 50 7,749, , ,112, ,058, Not Available 145 2,523, , Median Income < $25, ,504, , ,174, , >=$25,000 to <$30, ,917, , ,714, ,678, >=$30,000 to <$35,000 1,372 22,970, ,554, ,417, ,448, >=$35,000 to <$40,000 1,215 40,443, ,856, ,707, ,768, >=$40,000 to <$45, ,780, ,571, ,201, ,971, >=$45,000 to <$50, ,218, ,410, ,042, ,371, >=$50,000 1,180 40,936, ,763, ,941, ,221, Not Available 151 2,492, High School Education < 60% 126 1,817, , ,045, , >=60% to <70% ,944, , ,476, ,804, >=70% to <80% 1,646 49,285, ,466, ,908, ,632, >=80% to <90% 3,111 93,198, ,509, , ,753,772 1, ,221, >=90% ,495, ,143, ,984, ,002, Not Available 151 2,523, , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 8

180 EAC Election Day Survey Voting Equipment Usage 2004 General Election Updated: 09/19/ :06:28 Voting Equipment Usage Punchcard Voting Equipment (Inc.Datavote) Optical Scan Voting Equipment Electronic Voting Equipment Election 2004 Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Administration Total of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Code Name Jurisdictions Registration Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using Political Battleground States in 2004 Presidential Election Yes 3,093 64,166, ,057, , ,294,051 1, ,715, No 3, ,098, ,709, , ,904,577 1, ,352, Margin of Victory in 2004 Presidential Election < 2.5% ,923, ,217, ,963, ,946, >=2.5% to < 5.0% ,133, ,232, ,717, ,261, >=5.0% to < 7.5% ,830, ,110, ,074, ,246, >=7.5% to < 10.0 % 429 8,833, , ,090, ,396, >=10.0 % 4, ,044, ,519, , ,352,770 1, ,217, Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,178, ,275, , ,438,890 1, ,299, Bush 50% to 55% ,682, ,675, ,989, ,788, Bush < 50% 136 2,041, , , , Kerry < 50% 150 4,850, , ,213, , Kerry 50% to 55% ,160, ,261, ,455, ,507, Kerry > 55% 1,161 49,846, ,212, ,763, ,467, Tied 25 14, , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 8

181 EAC Election Day Survey Voting Equipment Usage 2004 General Ele Updated: 09/19/ :06:28 Election 2004 Administration Total Code Name Jurisdictions Registration 01 Alabama 67 2,597, Alaska 1 472, Arizona 15 2,642, Arkansas 75 1,699, California 58 16,646, Colorado 64 3,101, Connecticut 169 1,831, Delaware 3 553, District of Columbia 1 383, Florida 67 10,300, Georgia 159 4,248, Hawaii 5 647, Idaho , Illinois 110 7,195, Indiana 92 4,296, Iowa 99 2,226, Kansas 105 1,695, Kentucky 120 2,794, Louisiana 64 2,932, Maine 517 1,026, Maryland 24 3,105, Massachusetts 351 4,098, Michigan 83 7,164, Minnesota 87 2,977, Mississippi 82 1,469, Missouri 116 4,194, Montana , Nebraska 93 1,160, Nevada 17 1,073, New Hampshire , New Jersey 21 5,011, New Mexico , New York 58 11,837, North Carolina 100 5,526, North Dakota , Ohio 88 7,965, Oklahoma 77 2,143, Oregon 36 2,141, Pennsylvania 67 8,366, Rhode Island , South Carolina 46 2,318, South Dakota , Tennessee 95 3,748, Texas ,098, Utah 29 1,278, Vermont , Virginia 134 4,515, Washington 39 3,508, West Virginia 55 1,168, Wisconsin 1,910 4,179, Wyoming , American Samoa 1 66 Guam 1 72 Puerto Rico 110 2,440, Virgin Islands 1 50,731 Total 6, ,265,030 Maximum 1,910 16,646,555 Average 119 3,344,623 Minimum 1 50,731 Lever Machine Voting Equipment Voting Equipment Usage Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Cases = Number of Jurisdictions Reporting Subject Matter of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Number % of Regis in % of Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using of Juris. Juris. Juris. Cases Regis , , , , , ,349, ,831, , , , , , , , , , , ,329, , , , , , ,593, , , , , , , ,302, Paper Ballots Voting Equipment Mixed Voting Equipment , ,837, , , , , ,022, , , Unknown (Not Reported) , ,214, , , , , , ,317, ,264, , , , ,940, , , , , , , , , , ,440, , ,662, , ,085,167 1, ,997, ,484, ,837, , ,593, ,214, ,203, , , ,316, , , , Election Data Services, Inc. Confidential 9/22/05 Page 5 of 8

182 EAC Election Day Survey Voting Equipment Usage 2004 General Ele Updated: 09/19/ :06:28 Election 2004 Administration Total Code Name Jurisdictions Registration Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,484,493 Punch card ,767,547 Lever ,662,619 Paper 1,734 3,085,167 Optical scan 2,541 69,198,628 Electronic ,068,685 Multiple Systems ,997,891 Changed Voting Equipment Since 2000 General Election Yes 1,753 51,149,755 No 4, ,115,275 State Wide Voter Registration System in Place Yes 1,335 37,384,852 No 5, ,880,178 Election Day Registration Yes 2,823 10,323,368 No 3, ,941,662 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,077,741 In Precinct Only 4, ,336,604 None 1,056 8,850,685 No Excuse Absentee Balloting Yes 3,781 64,333,790 No 2, ,931,240 Early Voting Allowed Yes 1,701 73,710,075 No 4, ,554,955 Covered By Section 203, Language Minority Requirements Yes ,756,496 No 6, ,508,534 Covered By Section 5 of Voting Rights Act Yes ,868,855 No 5, ,396,175 Lever Machine Voting Equipment Voting Equipment Usage Paper Ballots Voting Equipment Mixed Voting Equipment Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Cases = Number of Jurisdictions Reporting Subject Matter Unknown (Not Reported) of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Number % of Regis in % of Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using of Juris. Juris. Juris. Cases Regis ,662, , ,085,167 1, ,997, ,484, , , ,006, ,138, ,209, , ,622,448 1, ,991, ,346, ,279, , ,207, , ,383, , ,836,826 1, ,790, ,852, , , ,192,850 1, , , ,548, ,892, ,407, ,612, ,499, , ,183, , ,162, ,047, ,266, ,476, , , ,502, ,507, , ,317,270 1, ,570, ,378, ,155, ,767, ,427, ,106, , , ,047, ,770, ,110, , ,235,842 1, ,950, ,713, ,456, , ,036, ,769, ,206, , ,711,026 1, ,961, ,714, ,195, , ,302, ,273, ,466, , ,735,032 1, ,695, ,210, Election Data Services, Inc. Confidential 9/22/05 Page 6 of 8

183 EAC Election Day Survey Voting Equipment Usage 2004 General Ele Updated: 09/19/ :06:28 Election 2004 Administration Total Code Name Jurisdictions Registration Demographics Region Northeast 1,710 34,273,670 South 1,423 62,606,676 Midwest 2,902 44,048,138 West ,845,684 Territories 113 2,490,862 Urban to Rural Urban ,441,314 Suburban ,552,530 Small Towns 1,710 44,193,768 Rural 3,307 19,586,556 Not Available - Territories 113 2,490,862 Size of Jurisdiction (VAP) < 1,000 1, ,006 >=1,000 to <3,500 1,165 2,182,148 >=3,500 to <10,000 1,043 5,966,645 >=10,000 to <50,000 1,704 31,472,681 >=50,000 to <250, ,992,270 >=250,000 to <1,000, ,396,493 >=1,000, ,867,508 Not Available 144 2,492,279 Race and Ethnicity Predominently NH White 6, ,662,585 Predominently NH Black 85 3,098,023 Predominently NH Native American ,022 Predominently Hispanic 50 7,749,995 Not Available 145 2,523,405 Median Income < $25, ,504,552 >=$25,000 to <$30, ,917,739 >=$30,000 to <$35,000 1,372 22,970,583 >=$35,000 to <$40,000 1,215 40,443,694 >=$40,000 to <$45, ,780,840 >=$45,000 to <$50, ,218,675 >=$50,000 1,180 40,936,586 Not Available 151 2,492,361 High School Education < 60% 126 1,817,027 >=60% to <70% ,944,978 >=70% to <80% 1,646 49,285,773 >=80% to <90% 3,111 93,198,279 >=90% ,495,512 Not Available 151 2,523,461 Lever Machine Voting Equipment Voting Equipment Usage Paper Ballots Voting Equipment Mixed Voting Equipment Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Cases = Number of Jurisdictions Reporting Subject Matter Unknown (Not Reported) of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Number % of Regis in % of Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using of Juris. Juris. Juris. Cases Regis ,993, , , , ,489, , ,572, ,273, , ,520, ,162, ,075, , , ,164, , ,490, ,240, , ,518, ,829, ,290, , ,324, ,011, ,144, , ,233, ,456, , , ,525,718 1, , , ,490, , , ,821 1, , , , , , , , , , ,314, , ,224, ,832, ,074, ,231, ,783, ,417, , ,038, ,863, ,342, ,416, ,856, ,491, ,472, , ,058,990 1, ,472, ,476, , , , , , , ,463, ,491, , , , , , , , , ,964, , , , ,517, , ,891, ,354, ,831, , ,457, ,505, ,569, , ,381, ,281, ,613, ,044, ,150, ,201, ,491, , , , , ,181, , , , ,525, , ,380, ,744, ,092, ,891, ,791, ,938, , , ,520, , ,491, Election Data Services, Inc. Confidential 9/22/05 Page 7 of 8

184 EAC Election Day Survey Voting Equipment Usage 2004 General Ele Updated: 09/19/ :06:28 Election 2004 Administration Total Code Name Jurisdictions Registration Political Battleground States in 2004 Presidential Election Yes 3,093 64,166,639 No 3, ,098,391 Margin of Victory in 2004 Presidential Election < 2.5% ,923,548 >=2.5% to < 5.0% ,133,130 >=5.0% to < 7.5% ,830,932 >=7.5% to < 10.0 % 429 8,833,490 >=10.0 % 4, ,044,988 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,178,580 Bush 50% to 55% ,682,203 Bush < 50% 136 2,041,746 Kerry < 50% 150 4,850,492 Kerry 50% to 55% ,160,396 Kerry > 55% 1,161 49,846,628 Tied 25 14,032 Lever Machine Voting Equipment Voting Equipment Usage Paper Ballots Voting Equipment Mixed Voting Equipment Number % of Regis in % of Number % of Regis in % of Number % of Regis in % of Cases = Number of Jurisdictions Reporting Subject Matter Unknown (Not Reported) of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis of Juris. Juris. Juris. Regis Number % of Regis in % of Using Using Using Cases Using Using Using Using Cases Using Using Using Using Cases Using of Juris. Juris. Juris. Cases Regis ,356, ,804, ,357, ,580, ,306, ,280, ,640, ,903, ,180, , ,878, , ,049, , ,109, , ,725, , , , , , , , ,715, , ,615,393 1, ,237, ,387, ,075, ,168, ,490, ,429, ,187, , ,007, ,632, , , , , ,418, , , , ,135, ,022, ,905, ,873, ,347, , ,640, ,023, , Election Data Services, Inc. Confidential 9/22/05 Page 8 of 8

185 Chapter 10 Voting Machines Table 10 presents the results of the Election Day Survey on numbers of voting machines and methods for tabulating ballots. Information on ballet tabulation methods i.e., whether ballots are counted in each precinct or at a central site was not requested by the survey, but was included by several states in their responses to question 13. For the other states, the appropriate ballot-tabulation method was determined from information about the voting equipment hardware. For example, if a jurisdiction specified ES&S Model 100 as the type of voting equipment used, information from the manufacturer was used to determine that ES&S Model 100 is a precinct-based, voter-activated paper ballot counter and vote tabulator. 1 The importance of where ballots are counted surfaced after the 2000 Florida debacle when studies, including those done by Election Data Services, found that voting systems that required the ballots be taken to a central location to be tallied had nearly five times the error rate as the same system that allowed the ballots to be counted in the precinct. This is because in-precinct tallying systems allowed the voter to feed his ballot into the machine and immediately be notified if he had overvoted an office. Since the ballot was kicked out of the tallying system immediately, this allowed a voter to correct his ballot and resubmit it. As a result, error rates dropped fivefold for in-precinct tallying systems. The only voting methods that allow an overvote to occur, and for which in-precinct-tallying systems would help prevent, are punch cards, paper ballots, and optical scan ballots. If programmed and set up correctly, lever machines and electronic voting systems have safeguards that are designed to prevent a voter from invalidating their ballot through an overvote. As noted above, Election Data Services attempted to code the tallying process for punch card and optical scan systems as either inprecinct or central count, depending upon the actual model of voting system used, if reported. Since in-precinct or central tallying was not asked by the U.S. Election Assistance Commission (EAC), it is impossible to code paper ballot jurisdictions similarly. Applicability and Coverage Only about 20 states provided information on the number of voting machines in use. The following states did not provide any information on the number of units used in a jurisdiction, which made it impossible to perform all the calculations initially intended for this chapter: Alabama, Colorado, Connecticut, District of Columbia, Georgia, Indiana, Maine, Minnesota, Missouri, New Hampshire, North Dakota, Rhode Island, South Carolina, South Dakota, Tennessee, Utah, Vermont, Washington, and Wisconsin. Information on the process for ballot tabulation is complete for only nine states. In the survey follow-up review, state election directors were asked to update the inprecinct versus central-count information for voting equipment in use, but only a few states provided this information. 1 Election Systems & Software, Product Overview, Model 100 Precinct Ballot Counter (Omaha, Neb.: nd).

186 2004 Election Day Survey Report, Part 2 Survey Results Voting Machines, Page 10-2 September 27, 2005 The lack of information on number of units prevents any calculations from being made showing the average number of units per precinct and/or polling place. Misunderstandings about what was asked for also caused unusual numbers to be calculated. For example, in Illinois, the state asked the counties for the Number of units of voting equipment (tabulators) and has the voting system previously been used in a Federal election while the EAC defined its request for the number of units used in the county (and/or precinct, if available). The intent of the question for punch card systems, and the responses from most states, dealt with the number of actual voting devices used by the county, not the number of counters used. If a responding jurisdiction assumed number of counters was the question, then the calculation of average number of units per precinct or polling place would quite naturally drop below one per precinct. On the other hand, for optical scan systems, it is quite possible that jurisdictions looked at the tabulators as the number to provide, and therefore, there are far more instances of fewer than one device per precinct. Electronic and lever machines do not present this apparent confusion, and therefore average number of units per precinct fall into an acceptable and understandable range. This chapter was intended to seek to answer the question of whether long lines might have been caused by too few polling booths. However, because of the survey confusion, answers are not possible at this time. Survey Results Table 10 presents more detailed data on voting machines from question 13 on the Election Day Survey. The table contains several calculations involving the six different types of voting equipment (including multiple systems) based on usage by jurisdictions, precincts, and polling places. The column headings in Table 10 are as follows: Table 10 Column Headings. Voting Machines Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question Total Registration Number of active and inactive registered voters, number of persons who voted on Election Day in six states, and VAP data for North Dakota and jurisdictions in Wisconsin that do not have voter registration, from col. 4 of table 2 5 Precincts Number of precincts from survey question 19 6 Cases Number of jurisdictions that responded to question 19 7 Polling Places Number of polling places from survey question 20 8 Cases Number of jurisdictions that responded to question 20

187 2004 Election Day Survey Report, Part 2 Survey Results Voting Machines, Page 10-3 September 27, 2005 Table 10 Column Headings (cont.) Col. Heading Description 9 Number of Juris. Using Punch card Equipment 10 Number of Units, Punch card Number of jurisdictions that responded to survey question 13 and reported the use of punch cards Number of jurisdictions that responded to question 13 and reported the number of punch card units 11 Cases Number of jurisdictions that responded to question 13 and reported the use of punch cards and the number of units 12 Average # of Units per Precinct, Punch card Number of punch card units (col. 10) divided by the number of precincts (col. 5) 13 Cases Number of jurisdictions that responded to questions 13 and Average # of Units per Polling Place, Punch card Number of punch card units (col. 10) divided by the number of polling places (col. 7) 15 Cases Number of jurisdictions that responded to questions 13 and Number of Juris. with In-Precinct Counting, Punch card 17 Number of Juris. with Central Counting, Punch card 18 Number of Juris. Using Optical Scan Equipment 19 Number of Units, Optical Scan Number of jurisdictions that responded to question 13 and reported a precinct ballot tabulation method or other information from which the ballot tabulation method could be determined Number of jurisdictions that responded to question 13 and reported a central ballot tabulation method or other information from which the ballot tabulation method could be determined Number of jurisdictions that responded to survey question 13 and reported the use of optical scan equipment Number of jurisdictions that responded to question 13 and reported the number of optical scan units 20 Cases Number of jurisdictions that responded to question 13 and reported the use of optical scan equipment and the number of units 21 Average # of Units per Precinct, Optical Scan Number of optical scan units (col. 19) divided by the number of precincts (col. 5) 22 Cases Number of jurisdictions that responded to questions 13 and Average # of Units per Polling Place, Optical Scan Number of optical scan units (col. 19) divided by the number of polling places (col. 7) 24 Cases Number of jurisdictions that responded to questions 13 and Number of Juris. with In-Precinct Counting, Optical Scan 26 Number of Juris. with Central Counting, Optical Scan 27 Number of Juris. Using Electronic Equipment Number of jurisdictions that responded to question 13 and reported a precinct ballot tabulation method or other information from which the ballot tabulation method could be determined Number of jurisdictions that responded to question 13 and reported a central ballot tabulation method or other information from which the ballot tabulation method could be determined Number of jurisdictions that responded to survey question 13 and reported the use of electronic voting equipment

188 2004 Election Day Survey Report, Part 2 Survey Results Voting Machines, Page 10-4 September 27, 2005 Table 10 Column Headings (cont.) Col. Heading Description 28 Number of Units, Electronic Number of jurisdictions that responded to question 13 and reported the number of units of electronic voting equipment 29 Cases Number of jurisdictions that responded to survey question 13 and reported the use of electronic equipment and the number of units 30 Average # of Units per Precinct, Electronic Number of units of electronic voting equipment (col. 28) divided by the number of precincts (col. 5) 31 Cases Number of jurisdictions that responded to questions 13 and Average # of Units per Polling Place, Electronic Number of units of electronic voting equipment (col. 5) divided by the number of polling places (col. 7) 33 Cases Number of jurisdictions that responded to questions 13 and Number of Juris. Using Lever Machines 35 Number of Units, Lever Machines Number of jurisdictions that responded to survey question 13 and reported the use of mechanical lever machines Number of jurisdictions that responded to question 13 and reported the number of lever machines 36 Cases Number of jurisdictions that responded to survey question 13 and reported the use of and the number of lever machines 37 Average # of Units per Precinct, Lever Number of lever machines (col. 35) divided by the number of precincts (col. 5) 38 Cases Number of jurisdictions that responded to questions 13 and Average # of Units per Polling Place, Lever Number of lever machines (col. 35) divided by the number of polling places (col. 7) 40 Cases Number of jurisdictions that responded to questions 13 and Number of Juris. Using Paper ballots Number of Units, Paper ballots Cases Average # of Units per Precinct, Paper Ballots Number of jurisdictions that responded to survey question 13 and reported the use of paper ballots Number of jurisdictions that responded to question 13 and reported the number of paper ballots Number of jurisdictions that responded to survey question 13 and reported the use of and the number of paper ballots Number of paper ballots (col. 42) divided by the number of precincts (col. 5) 45 Cases Number of jurisdictions that responded to questions 13 and Average # of Units per Polling Place, Paper Number of paper ballots (col. 42) divided by the number of polling places (col. 7) 47 Cases Number of jurisdictions that responded to questions 13 and Number of Juris. Using Mixed equipment Number of Units, Mixed equipment Cases Number of jurisdictions that responded to survey question 13 and reported the use of mixed voting equipment Number of jurisdictions that responded to question 13 and reported the number of mixed voting equipment Number of jurisdictions that responded to survey question 13 and reported the use of and the number of mixed equipment

189 2004 Election Day Survey Report, Part 2 Survey Results Voting Machines, Page 10-5 September 27, 2005 Table 10 Column Headings (cont.) Col. Heading Description 51 Average # of Units per Precinct, Mixed Number of mixed equipment (col. 49) divided by the number of precincts (col. 5) 52 Cases Number of jurisdictions that responded to questions 13 and Average # of Units per Polling Place, Mixed Number of mixed equipment (col. 49) divided by the number of polling places (col. 7) 54 Cases Number of jurisdictions that responded to questions 13 and Number of Juris. Using Unknown equipment Number of Units, Unknown equipment Cases Average # of Units per Precinct, Unknown Number of jurisdictions that responded to survey question 13 and reported the use of unknown voting equipment Number of jurisdictions that responded to question 13 and reported the number of unknown voting equipment Number of jurisdictions that responded to survey question 13 and reported the use of and the number of unknown equipment Number of unknown equipment (col. 56) divided by the number of precincts (col. 5) 59 Cases Number of jurisdictions that responded to questions 13 and Average # of Units per Polling Place, Unknown Number of unknown equipment (col. 56) divided by the number of polling places (col. 7) 61 Cases Number of jurisdictions that responded to questions 13 and 20

190 2004 Election Day Survey Report, Part 2 Survey Results Voting Machines, Page 10-6 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 10 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 10 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary The lack of complete, or even near complete, data on machines and number of units makes any kind of meaningful analysis somewhat questionable. Only one-third of the punch card jurisdictions provided any data on number of units, and much of that data was probably incorrectly reported anyway. Fewer than half of the optical scan, lever, and electronic system jurisdictions reported the number of units used. With those limitations in mind, we have calculated that electronic system jurisdictions average three devices per precinct and slightly over five devices per polling place. The highest ratio of machines per location occurred in the state of Maryland, where between nine and 10 devices were used. Due to the high cost of lever machines, they average only 1.5 machines per precinct and slightly over three machines per polling place. The maximum number per precinct and polling place occurred in North Carolina and Wyoming, with slightly more than three and six per precinct, respectively. Of the data reported, 90 percent of the punch card jurisdictions in this country utilize a central-count tallying process, On the other hand, 61 percent of optical scan jurisdictions use a precinct-based tallying process. States See state information in summary above. Regions For optical scan usage, only jurisdictions in the West reported more central-count operations. All other regions used more precinct-count operations.

191 2004 Election Day Survey Report, Part 2 Survey Results Voting Machines, Page 10-7 September 27, 2005 Urban to Rural Rural areas of the nation that use optical scan systems are much more likely to use a central-count tallying process. Not surprisingly, these are some of the older models of optical scan systems, since the rural Midwest is where optical scan devices were first used as an easier way to tally paper ballots. Size of Jurisdiction The larger a jurisdiction the more likely they are to have more electronic or lever machines per polling location. The largest jurisdictions average four to six voting machines per polling location, but smaller jurisdictions average just two machines per location. Race and Ethnicity In those jurisdictions using optical scan systems, 62 percent of the predominantly White jurisdictions use precinct-based tallying systems. On the other hand 86 percent of the predominantly Hispanic jurisdictions using optical scan devices use the central-count tallying process. Predominantly African American jurisdictions that use optical scan systems are fairly evenly split between central- and precinct-tallying processes. Median Income In optical scan jurisdictions, the higher the median income of a community the less likely they are to use a centrally located tallying process. High School Education The higher the education levels, the more likely optical scan jurisdictions are using precinct-based tallying technology. Section 203 Language Minority Requirements Optical scan jurisdictions subject to Section 203 minority language requirements of the Voting Rights Act are over twice as likely to be operating a central-count tallying system. Section 5 Preclearance of Voting Procedures Optical scan jurisdictions subject to Section 5 of the Voting Rights Act are more likely to be operating a central-count tallying system, while non-section 5 jurisdictions are twice as likely to be tallying their ballots in the precinct. Type of Voting Equipment Not Applicable. Changed Voting Equipment since 2000 Jurisdictions that changed their voting system in the last four years and went to the optical scan system are more than four times as likely to adopt a precinct-based tallying system. Statewide Voter Registration Database No real patterns are discernible in this subcategory.

192 2004 Election Day Survey Report, Part 2 Survey Results Voting Machines, Page 10-8 September 27, 2005 Election Day Registration No real patterns are discernible in this subcategory. Provisional Ballot Acceptance No real patterns are discernible in this subcategory. No Excuse Absentee Balloting No real patterns are discernible in this subcategory. Early Voting No real patterns are discernible in this subcategory. Battleground States Perhaps anticipating the increased turnout in battleground states, all voting system jurisdictions showed a larger number of units per polling location than in nonbattleground states. Presidential Margin of Victory For both punch card and electronic voting jurisdictions, there appears to be a slight trend toward more machines per polling location the closer the margin of victory in the 2004 presidential election. Red versus Blue Jurisdictions No real patterns are discernible in this subcategory.

193 EAC Election Day Survey Voting Machines 2004 General Election Voting Machines Updated: 09/19/ :07:31 Punchcard Voting Equipment (including Datavote) Optical Scan Voting Equipment Election 2004 Number Number Average # of Units Number of Juris with Number Number Average # of Units Number of Juris with Administration Total Polling of Juris. of Per Per In-Precinct Central of Juris. of Per Per In-Precinct Central Code Name Jurisdictions Registration Precincts Cases Places Cases Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting 01 Alabama 67 2,597,629 2, , Alaska 1 472, Arizona 15 2,642,120 2, , , Arkansas 75 1,699,934 2, , California 58 16,646,555 21, , , , Colorado 64 3,101,956 3, , Connecticut 169 1,831, Delaware 3 553, District of Columbia 1 383, Florida 67 10,300,942 6, , , Georgia 159 4,248,802 3, , Hawaii 5 647, Idaho , , Illinois 110 7,195,882 11, , , , Indiana 92 4,296,602 5, , Iowa 99 2,226,721 1, , , Kansas 105 1,695,457 3, , Kentucky 120 2,794,286 3, , , Louisiana 64 2,932,142 4, , Maine 517 1,026, Maryland 24 3,105,370 1, , Massachusetts 351 4,098,634 2, , , Michigan 83 7,164,047 5, , , Minnesota 87 2,977,496 4, Mississippi 82 1,469,608 1, , Missouri 116 4,194,416 5, , Montana , , Nebraska 93 1,160,193 1, , Nevada 17 1,073,869 1, New Hampshire , New Jersey 21 5,011,693 6, , New Mexico , New York 58 11,837,068 15, , North Carolina 100 5,526,981 2, , , , North Dakota , Ohio 88 7,965,110 11, , , Oklahoma 77 2,143,978 2, , , Oregon 36 2,141,249 1, Pennsylvania 67 8,366, , Rhode Island , South Carolina 46 2,318,235 2, South Dakota , Tennessee 95 3,748,235 2, , Texas ,098,329 8, , , , Utah 29 1,278,912 1, , Vermont , Virginia 134 4,515,675 2, , , , Washington 39 3,508,208 6, , West Virginia 55 1,168,694 1, , Wisconsin 1,910 4,179,774 3,563 1,253 2,686 1, Wyoming , American Samoa 1 66 Guam 1 72 Puerto Rico 110 2,440,131 1, , Virgin Islands 1 50, Total 6, ,265, ,252 5, ,754 5, , ,541 37, Maximum 1,910 16,646,555 21,857 1,253 14,467 1, , , Average 119 3,344,623 3, , , , Minimum 1 50, Election Data Services, Inc. Confidential 9/22/05 Page 1 of 12

194 EAC Election Day Survey Voting Machines 2004 General Election Voting Machines Updated: 09/19/ :07:31 Punchcard Voting Equipment (including Datavote) Optical Scan Voting Equipment Election 2004 Number Number Average # of Units Number of Juris with Number Number Average # of Units Number of Juris with Administration Total Polling of Juris. of Per Per In-Precinct Central of Juris. of Per Per In-Precinct Central Code Name Jurisdictions Registration Precincts Cases Places Cases Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting Election Administration Voting Equipment Used in 2004 General Election None / Unknown ,484,493 13, , Punch card ,767,547 19, , , Lever ,662,619 20, , Paper 1,734 3,085,167 5,704 1,573 3,416 1,183 Optical scan 2,541 69,198,628 69,370 2,405 46,265 2,185 2,541 37, Electronic ,068,685 35, , Multiple Systems ,997,891 10, , Changed Voting Equipment Since 2000 General Election Yes 1,753 51,149,755 46,241 1,296 31,649 1, , No 4, ,115, ,011 4,100 82,105 3, , ,917 26, State Wide Voter Registration System in Place Yes 1,335 37,384,852 33,575 1,153 20,815 1, , , No 5, ,880, ,677 4,243 92,939 4, , ,986 26, Election Day Registration Yes 2,823 10,323,368 9,704 1,924 3,794 1, , No 3, ,941, ,548 3, ,960 3, , ,693 36, Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,077,741 65,986 1,080 44,212 1, , , In Precinct Only 4, ,336, ,295 3,504 66,513 3, , ,758 23, None 1,056 8,850,685 7, , , No Excuse Absentee Balloting Yes 3,781 64,333,790 70,535 3,106 47,225 2, , ,408 23, No 2, ,931, ,717 2,290 66,529 2, , ,133 13, Early Voting Allowed Yes 1,701 73,710,075 69,882 1,683 51,609 1, , , No 4, ,554, ,370 3,713 62,145 3, , ,696 12, Covered By Section 203, Language Minority Requirements Yes ,756,496 54, , , , No 6, ,508, ,201 4,953 77,656 4, , ,290 26, Covered By Section 5 of Voting Rights Act Yes ,868,855 32, , , , No 5, ,396, ,276 4,541 88,074 4, , ,188 30, Election Data Services, Inc. Confidential 9/22/05 Page 2 of 12

195 EAC Election Day Survey Voting Machines 2004 General Election Voting Machines Updated: 09/19/ :07:31 Punchcard Voting Equipment (including Datavote) Optical Scan Voting Equipment Election 2004 Number Number Average # of Units Number of Juris with Number Number Average # of Units Number of Juris with Administration Total Polling of Juris. of Per Per In-Precinct Central of Juris. of Per Per In-Precinct Central Code Name Jurisdictions Registration Precincts Cases Places Cases Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting Demographics Region Northeast 1,710 34,273,670 25,068 1,230 13, , , South 1,423 62,606,676 48,810 1,408 37,805 1, , , Midwest 2,902 44,048,138 55,993 2,243 35,954 2, , ,135 7, West ,845,684 42, , , , Territories 113 2,490,862 1, , Urban to Rural Urban ,441,314 60, , , , Suburban ,552,530 37, , , , Small Towns 1,710 44,193,768 41,994 1,421 28,085 1, , , Rural 3,307 19,586,556 32,252 2,780 21,938 2, , ,133 10, Not Available - Territories 113 2,490,862 1, , Size of Jurisdiction (VAP) < 1,000 1, ,006 2,118 1,229 1,350 1, >=1,000 to <3,500 1,165 2,182,148 2, , >=3,500 to <10,000 1,043 5,966,645 8, , , >=10,000 to <50,000 1,704 31,472,681 35,443 1,554 25,830 1, , , >=50,000 to <250, ,992,270 41, , , , >=250,000 to <1,000, ,396,493 44, , , , >=1,000, ,867,508 38, , , , Not Available 144 2,492,279 1, , Race and Ethnicity Predominently NH White 6, ,662, ,698 5, ,108 4, , ,469 36, Predominently NH Black 85 3,098,023 2, , Predominently NH Native American , Predominently Hispanic 50 7,749,995 7, , Not Available 145 2,523,405 1, , Median Income < $25, ,504,552 3, , >=$25,000 to <$30, ,917,739 12, , , >=$30,000 to <$35,000 1,372 22,970,583 23,424 1,197 16,639 1, , , >=$35,000 to <$40,000 1,215 40,443,694 40,250 1,056 24, , , >=$40,000 to <$45, ,780,840 36, , , , >=$45,000 to <$50, ,218,675 19, , , , >=$50,000 1,180 40,936,586 36, , , , Not Available 151 2,492,361 1, , High School Education < 60% 126 1,817,027 2, , >=60% to <70% ,944,978 18, , , >=70% to <80% 1,646 49,285,773 51,393 1,411 32,782 1, , , >=80% to <90% 3,111 93,198,279 87,644 2,502 56,581 2, , ,243 23, >=90% ,495,512 13, , , , Not Available 151 2,523,461 1, , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 12

196 EAC Election Day Survey Voting Machines 2004 General Election Voting Machines Updated: 09/19/ :07:31 Punchcard Voting Equipment (including Datavote) Optical Scan Voting Equipment Election 2004 Number Number Average # of Units Number of Juris with Number Number Average # of Units Number of Juris with Administration Total Polling of Juris. of Per Per In-Precinct Central of Juris. of Per Per In-Precinct Central Code Name Jurisdictions Registration Precincts Cases Places Cases Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting Using Units Cases Precinct Cases Polling Pl. Cases Counting Counting Political Battleground States in 2004 Presidential Election Yes 3,093 64,166,639 59,123 2,113 33,037 2, , ,172 9, No 3, ,098, ,129 3,283 80,717 2, , ,369 28, Margin of Victory in 2004 Presidential Election < 2.5% ,923,548 13, , , , >=2.5% to < 5.0% ,133,130 10, , , , >=5.0% to < 7.5% ,830,932 13, , , , >=7.5% to < 10.0 % 429 8,833,490 9, , , , >=10.0 % 4, ,044, ,787 3,788 83,067 3, , ,846 24, Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,178,580 68,994 2,690 49,173 2, , ,306 17, Bush 50% to 55% ,682,203 25, , , , Bush < 50% 136 2,041,746 1, , Kerry < 50% 150 4,850,492 4, , , Kerry 50% to 55% ,160,396 22, , , , Kerry > 55% 1,161 49,846,628 49, , , , Tied 25 14, Election Data Services, Inc. Confidential 9/22/05 Page 4 of 12

197 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases 01 Alabama Alaska 1 04 Arizona Arkansas California Colorado Connecticut Delaware 3 11 District of Columbia 1 12 Florida Georgia Hawaii 5 16 Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin 1, Wyoming American Samoa 1 66 Guam 1 72 Puerto Rico Virgin Islands 1 Total 6,568 Maximum 1,910 Average 119 Minimum 1 Electronic Voting Equipment Lever Machine Voting Equipment Paper Ballots Voting Equipment , , , , , , , , , , , , , , , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 5 of 12

198 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Election Administration Voting Equipment Used in 2004 General Election None / Unknown 908 Punch card 260 Lever 394 Paper 1,734 Optical scan 2,541 Electronic 608 Multiple Systems 123 Changed Voting Equipment Since 2000 General Election Yes 1,753 No 4,815 State Wide Voter Registration System in Place Yes 1,335 No 5,233 Election Day Registration Yes 2,823 No 3,745 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 In Precinct Only 4,350 None 1,056 No Excuse Absentee Balloting Yes 3,781 No 2,787 Early Voting Allowed Yes 1,701 No 4,867 Covered By Section 203, Language Minority Requirements Yes 468 No 6,100 Covered By Section 5 of Voting Rights Act Yes 880 No 5,688 Electronic Voting Equipment Lever Machine Voting Equipment Paper Ballots Voting Equipment , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 6 of 12

199 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Demographics Region Northeast 1,710 South 1,423 Midwest 2,902 West 420 Territories 113 Urban to Rural Urban 567 Suburban 871 Small Towns 1,710 Rural 3,307 Not Available - Territories 113 Size of Jurisdiction (VAP) < 1,000 1,761 >=1,000 to <3,500 1,165 >=3,500 to <10,000 1,043 >=10,000 to <50,000 1,704 >=50,000 to <250, >=250,000 to <1,000, >=1,000, Not Available 144 Race and Ethnicity Predominently NH White 6,264 Predominently NH Black 85 Predominently NH Native American 24 Predominently Hispanic 50 Not Available 145 Median Income < $25, >=$25,000 to <$30, >=$30,000 to <$35,000 1,372 >=$35,000 to <$40,000 1,215 >=$40,000 to <$45, >=$45,000 to <$50, >=$50,000 1,180 Not Available 151 High School Education < 60% 126 >=60% to <70% 661 >=70% to <80% 1,646 >=80% to <90% 3,111 >=90% 873 Not Available 151 Electronic Voting Equipment Lever Machine Voting Equipment Paper Ballots Voting Equipment 23 9, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 7 of 12

200 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Political Battleground States in 2004 Presidential Election Yes 3,093 No 3,475 Margin of Victory in 2004 Presidential Election < 2.5% 515 >=2.5% to < 5.0% 476 >=5.0% to < 7.5% 510 >=7.5% to < 10.0 % 429 >=10.0 % 4,492 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 Bush 50% to 55% 982 Bush < 50% 136 Kerry < 50% 150 Kerry 50% to 55% 872 Kerry > 55% 1,161 Tied 25 Electronic Voting Equipment Lever Machine Voting Equipment Paper Ballots Voting Equipment 66 39, , , , , , , , , , , , , , , , , , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 8 of 12

201 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases 01 Alabama Alaska 1 04 Arizona Arkansas California Colorado Connecticut Delaware 3 11 District of Columbia 1 12 Florida Georgia Hawaii 5 16 Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin 1, Wyoming American Samoa 1 66 Guam 1 72 Puerto Rico Virgin Islands 1 Total 6,568 Maximum 1,910 Average 119 Minimum , , Mixed Voting Equipment Cases = Number of Jurisdictions Reporting Subject Matter , , , Unknown / Not Reported , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 9 of 12

202 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Election Administration Voting Equipment Used in 2004 General Election None / Unknown 908 Punch card 260 Lever 394 Paper 1,734 Optical scan 2,541 Electronic 608 Multiple Systems 123 Changed Voting Equipment Since 2000 General Election Yes 1,753 No 4,815 State Wide Voter Registration System in Place Yes 1,335 No 5,233 Election Day Registration Yes 2,823 No 3,745 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 In Precinct Only 4,350 None 1,056 No Excuse Absentee Balloting Yes 3,781 No 2,787 Early Voting Allowed Yes 1,701 No 4,867 Covered By Section 203, Language Minority Requirements Yes 468 No 6,100 Covered By Section 5 of Voting Rights Act Yes 880 No 5,688 Mixed Voting Equipment , Cases = Number of Jurisdictions Reporting Subject Matter Unknown / Not Reported , , , , , , , , , , , , , , , , , , , , , , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 10 of 12

203 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Demographics Region Northeast 1,710 South 1,423 Midwest 2,902 West 420 Territories 113 Urban to Rural Urban 567 Suburban 871 Small Towns 1,710 Rural 3,307 Not Available - Territories 113 Size of Jurisdiction (VAP) < 1,000 1,761 >=1,000 to <3,500 1,165 >=3,500 to <10,000 1,043 >=10,000 to <50,000 1,704 >=50,000 to <250, >=250,000 to <1,000, >=1,000, Not Available 144 Race and Ethnicity Predominently NH White 6,264 Predominently NH Black 85 Predominently NH Native American 24 Predominently Hispanic 50 Not Available 145 Median Income < $25, >=$25,000 to <$30, >=$30,000 to <$35,000 1,372 >=$35,000 to <$40,000 1,215 >=$40,000 to <$45, >=$45,000 to <$50, >=$50,000 1,180 Not Available 151 High School Education < 60% 126 >=60% to <70% 661 >=70% to <80% 1,646 >=80% to <90% 3,111 >=90% 873 Not Available 151 Mixed Voting Equipment Cases = Number of Jurisdictions Reporting Subject Matter Unknown / Not Reported , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 11 of 12

204 EAC Election Day Survey Voting Machines 2004 General Ele Voting Machines Updated: 09/19/ :07:31 Election Number Number Average # of Units Number Number Average # of Units Administration of Juris. of Per Per of Juris. of Per Per Code Name Jurisdictions Using Units Cases Precinct Cases Polling Pl. Cases Using Units Cases Precinct Cases Polling Pl. Cases Political Battleground States in 2004 Presidential Election Yes 3,093 No 3,475 Margin of Victory in 2004 Presidential Election < 2.5% 515 >=2.5% to < 5.0% 476 >=5.0% to < 7.5% 510 >=7.5% to < 10.0 % 429 >=10.0 % 4,492 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 Bush 50% to 55% 982 Bush < 50% 136 Kerry < 50% 150 Kerry 50% to 55% 872 Kerry > 55% 1,161 Tied 25 Mixed Voting Equipment Cases = Number of Jurisdictions Reporting Subject Matter Unknown / Not Reported 62 4, , , , , , , , , , , , , , , , , , , , Election Data Services, Inc. Confidential 9/22/05 Page 12 of 12

205 Chapter 11 Voting Equipment Malfunctions The U.S. Election Assistance Commission (EAC) sought information in the Election Day Survey about voting equipment malfunctions that occurred at the November 2004 general election. The survey asked state election directors to identify by county and precinct, if available, where any of the following malfunctions occurred and whether the affected voting machines were returned to service: 14a. Power failure An interruption or failure that renders a voting machine incapable of counting votes for more than five minutes during Election Day, absentee, or early voting. 14b. Broken counter A malfunction of a lever voting machine that renders the machine incapable of counting additional votes on any votable position. 14c. Computer failure A software, hardware, or firmware malfunction, disablement, or interruption that renders a voting machine incapable of presenting the ballot, recording votes, or printing or tabulating results. 14d. Printer failure A malfunction or interruption of the printer hardware, software, or mechanical components constituting the mechanism for creating a printed result of all contests voted (includes printers on electronic and mechanical lever voting machines). 14e. Screen failure A malfunction or interruption of the screen display or indicator lights on an electronic voting machine (DRE) that renders the machine incapable of indicating which choices a voter has made or which races for which the voter is eligible to vote. 14f. Fatal damage to machine Damage to or destruction of a voting machine that renders the machine incapable of recording votes or printing the results of voting. 14g. Modem failure A malfunction or interruption of a modem or the computer hardware or software using a modem to transmit election results to a central-counting location that renders the modem incapable of transmitting election results. 14h. Scanner failure A malfunction or interruption of a paper ballot reading device that either renders the device incapable of counting votes or renders the tabulated results inaccurate. 14i. Ballot encoder/activator failure A malfunction or interruption of that piece of a DRE that encodes a smart card or other similar device with the voter s ballot or with critical demographic data that allows the machine to access the proper ballot for the voter. 14j. Audio ballot failure A malfunction or interruption of the hardware, software, or peripherals that renders a DRE incapable of playing an audio version of the ballot. 14k. Other voting machine malfunctions

206 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Malfunctions, Page 11-2 September 27, 2005 Applicability and Coverage Question 14 on voting equipment malfunctions had by far the least coverage of any of the survey questions. Twenty-one states did not respond to the question or said that information on malfunctions was not available. Two states said that no voting equipment malfunctions were reported, and 10 states reported only one to six malfunctions statewide. Of 20 states that provided detailed county-by-county responses, only eight specified whether the affected machines were returned to service and only four identified the precincts in which the malfunctions occurred. In all, we have information from only 485 of the 6,567 jurisdictions in the EAC database, or only seven percent of the country. Table 14a shows the type of voting equipment used by these jurisdictions. Of these 485 jurisdictions, 210 used optical scan systems, 191 used electronic voting systems, 36 used lever machines, 11 used punch cards, and four used hand-counted paper ballots. Thirteen responses were from jurisdictions using more than one type of voting equipment. Figure 11.1 Jurisdictions with Equipment Malfunctions Punch Card, 11, 2% Mixed, 13, 3% Paper ballots, 4, 1% Electronic, 191, 41% Optical Scan, 210, 45% Lever, 36, 8% Electronic Lever Optical Scan Paper ballots Punch Card Mixed Survey Results Due to the small number of responses to the question on voting equipment malfunctions, we did not create the standard data table that forms the basis for other chapters in this report. Instead, we created a table that shows the number of malfunctions reported for each voting equipment type. That information is in Table 11a, along with the number of jurisdictions for each equipment type.

207 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Malfunctions, Page 11-3 September 27, 2005 Table 11a. Reported Voting Equipment Malfunctions by Equipment Type Total Number Malfunctions Electronic Lever Optical Scan Paper ballots Punch Cards Mixed (No. of Jurisdictions) (465) (191) (36) (210) (4) (11) (13) 14a. Power Failures b. Broken Counter c. Computer Failure d. Printer Failure e. Screen Failure f. Fatal Damage g. Modem Failure h. Scanner Failure i. Ballot Encoder/ Activator Failure j. Audio Ballot Failure k. Other Failure The responses from states to the Election Day Survey clearly show that certain types of equipment failures tend to be tied to certain types of equipment. Printer problems tend to be tied to electronic voting equipment, while scanner failures are most apt to occur with optical scan systems. Screen and power failures were mainly problems of electronic systems. Computer failures occurred in both electronic and optical scan systems. Although the EAC did not ask the reason for each voting equipment malfunction, except for the Other Failure category, some states provided that information for all categories. Tables 11b and 11c list the reasons provided by states for malfunctions affecting electronic and optical scan voting equipment. Table 11b. Reported Reasons for Electronic Voting Equipment Malfunctions Type of Malfunction Reason for Malfunction 14a. Power Failures Battery, Power Cord 14b. Broken Counter 14c. Computer Failure Export problem, Memory card 14d. Printer Failure Write-in door jam, Paper jam 14e. Screen Failure Blank screen, Frozen screen, Calibration problem 14f. Fatal Damage Motherboard 14g. Modem Failure 14h. Scanner Failure Feed problem 14i. Ballot Encoder/Activator Failure 14j. Audio Ballot Failure Jam, Incorrect ballot setup

208 2004 Election Day Survey Report, Part 2 Survey Results Voting Equipment Malfunctions, Page 11-4 September 27, 2005 Type of Malfunction Reason for Malfunction 14k. Other Failure Insufficient recycle time between voters, Curtain mechanism broken Table 11c. Reported Reasons for Optical Scan Voting Equipment Malfunctions Type of Malfunction Reason for Malfunction 14a. Power Failures Battery, Machine unplugged 14b. Broken Counter Unknown 14c. Computer Failure Memory pack, Memory card 14d. Printer Failure Ribbon malfunction, Paper jam 14e. Screen Failure 14f. Fatal Damage 14g. Modem Failure Telephone line 14h. Scanner Failure Ballot jam, Reader head, Channel coding problem, Tabulator problem 14i. Ballot Encoder/Activator Failure 14j. Audio Ballot Failure 14k. Other Failure Broken Counter, Pointer stuck, Underprinted ballots For lever machines, reasons for equipment malfunctions included stuck levers, incorrect ballot strips, broken curtain mechanisms, and printer failures (e.g., printer unplugged). For punch cards, reasons for equipment malfunctions included voter units not recognized (computer failure), data pack failures, broken counters, and ballot cards not punched through. REFERENCES Saltman, Roy Effective Use of Computing Technology in Vote-Tallying. Final project report prepared by Information Technology Division of Institute for Computer Sciences and Technology, National Bureau of Standards, for Clearinghouse on Election Administration, Office of Federal Elections, General Accounting Office. Washington, DC: GPO.

209 Chapter 12 Poll Workers Table 12 presents data from the Election Day Survey about poll workers. The survey asked about the number of poll workers statewide and in each local election jurisdiction, the required number of poll workers per precinct or polling place by law or regulation, the number of precincts or polling places in jurisdictions that did not have the required number of poll workers, and the number of additional poll workers that would have been needed to meet the staffing requirement in each precinct that had a deficit of poll workers. Poll workers were defined as persons who (a) verified the identity of a voter; (b) assisted the voter with signing the register, affidavits or other documents required to cast a ballot; (c) assisted the voter by providing the voter with a ballot or setting up the voting machine for the voter; or (d) served other functions dictated by state law on November 2, The definition excludes observers stationed at polling places. Applicability and Coverage All states have polling places and thus need poll workers. However, Oregon conducts all elections by mail, and locates one polling place in each county s administrative offices, and therefore does not have the same staffing requirements as other states that must staff polling places on Election Day. Historical Context The type of person who was considered to be a poll worker has changed over time. In the era of machine politics in the United States, poll workers were people selected and paid for by the political parties to attend to the passing out of party-printed ballots at the polling place (Sorauf 1954; Woodruff 1908). Little is mentioned about the poll workers who were responsible for collecting the ballots, though that, too, may have been the responsibility of the partisan-organized poll workers. A sort of checks and balances evolved, where both parties were responsible for stationing poll workers at every polling place to make sure the other side did not steal the election (Oestreicher 1988). During the turn of the last century, the Progressive movement initiated reforms designed to clean up the United States electoral system. The regulation of poll workers may have been one aspect of this reform effort; this was at least true in New Jersey (Lapomarda 1970). It was during this time that the modern image of the nonpartisan poll worker evolved. However, as a holdover from the machine era, many jurisdictions still require that the political parties nominate or provide lists of poll workers. Poll workers must be able to work on Election Day. Most jurisdictions require poll workers to work the entire day, while some arrange shifts. Duties range from managing the polling place, recording who votes on the registration list, registering voters to vote in states with Election Day registration, assisting voters in casting their vote, ensuring the voting equipment works properly, tallying the ballots (depending on the voting equipment in use), and transmitting information to the central-count location at the end of the day.

210 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-2 September 27, 2005 Most poll workers receive training in the elections process from local election administrators. Training for poll workers is documented as early as 1964 in Hamilton County, Ohio, where training was deemed necessary to learn how to use new voting machines (Willis 1966). In most cases, poll workers are compensated for their training time, but these rates vary greatly across the nation. While poll workers are often compensated for their time, being a poll worker is not a career. It is largely a voluntary activity. Near the close of the 1800s, one study documented that women served as poll workers prior to women s suffrage (Formisano 1999). Beginning in the late 1960s, as the female population went back into the workforce in greater numbers, the reservoir of available poll workers begin to dry up. Compensation pay was increased to reflect the scarcer commodity. Allowable precinct sizes were increased so that the total number of precincts could decrease to correspond with the smaller labor pool. Changing the type of voting equipment used in precincts was also an alternative used by election administrators to deal with the difficulty in finding poll workers. In modern times, the pool of potential volunteers consists of retired persons and college students. In addition, the U.S. Election Assistance Commission (EAC) has a special mandate under the Help America Vote Act of 2002 to encourage to college students to volunteer. Survey Results Table 12 presents data on poll workers from questions on the Election Day Survey. In the table, the average number of poll workers is calculated for precincts and polling places. The number of precincts with fewer than the required number of poll workers is calculated as a percentage of the total number of precincts. The column headings in Table 12 are as follows: Column Headings for Table 12. Poll Workers Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Number of Precincts Number of precincts from survey question 19 5 Cases Number of jurisdictions that responded to question 19 6 Total Number of Polling Places Number of polling places from survey question 20 7 Cases Number of jurisdictions that responded to question 20 8 Total Number of Poll Workers Number of poll workers from survey question 15 9 Cases Number of jurisdictions that responded to question Average # of Poll Workers per Precinct Number of poll workers (col. 8) divided by the number of precincts (col. 4) 11 Cases Number of jurisdictions that responded to question 15b on poll workers and question 19 on precincts

211 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-3 September 27, 2005 Column Headings for Table 12 (cont.) Col. Heading Description 12 Average # of Poll Workers Polling Place Number of poll workers (col. 8) divided by the number of polling places (col. 6) 13 Cases Number of jurisdictions that responded to questions 15 and Number of Precincts or Polling Places < Req. Poll Workers Number of precincts or polling places with fewer than the required number of poll workers from question 17a 15 Cases Number of jurisdictions that responded to question 17a 16 Percent Precincts < Req. Poll Workers Number of precincts with fewer than the required number of poll workers (col. 14) divided by total number of precincts (col. 4) 17 Cases Number of jurisdictions that responded to questions 4 and 17a 18 Cases > 100% 19 Percent Polling Places < Req. Poll Workers Number of cases where the reported number of precincts and polling places with fewer than the required number of poll workers (col. 16) is greater than the reported number of precincts (col. 4) Number of polling places with fewer than the required number of poll workers (col. 14) divided by total number of polling places (col. 6) 20 Cases Number of jurisdictions that responded to questions 17a and Cases > 100% Number of cases where the reported number of polling places with fewer than the required number of poll workers (col. 14) is greater than the number of polling places (col. 6)

212 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-4 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 12 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 12 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary In the analysis, we construct four measures from the responses to the Election Day Survey: the average number of poll workers per precinct, the average number of poll workers per polling place, the percentage of precincts reporting an insufficient number of poll workers, and the percentage of polling places reporting an insufficient number of poll workers. Nationally, jurisdictions reported an average of 7.9 poll workers per polling place and 5.7 poll workers per precinct. Jurisdictions reported that 5.8 percent of polling places and 4.0 percent of precincts did not have the minimum number of required poll workers. In all, 5,252 precincts or polling places of the 113,749 reported polling places or 174,252 reported precincts were said to have inadequate staffing. Generally, precincts and polling places are the same. An important qualification is that more than one precinct may be consolidated into one polling place, and consolidation occurs more often in urban jurisdictions, among others. (See polling place analysis in chapter 13 for further description and analysis.) States have different methods of defining polling places, and how they staff those locations affects the measurement of workers per polling place. Oregon, Wisconsin, and Puerto Rico report the number of poll workers per polling place in unique ways that confound the analysis: Oregon conducts elections by mail, and locates only one polling place in each county s administrative offices. In many instances, therefore, the count of poll workers represented the number of election staff workers within a county office.

213 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-5 September 27, 2005 Wisconsin administers elections within what they call wards, which may have created confusion with regard to how to report precincts and polling places on the Election Day Survey among jurisdictions within the state. For most jurisdictions, the reported number of polling places is much too low, e.g., six in the city of Milwaukee. In some cases, adding precincts and polling places appears to provide a reasonable number of polling places, e.g., Milwaukee reported 314 precincts. However, this decision rule is not consistent; Burlington City reported 34 polling places, 16 precincts, and 47 poll workers, which, if we sum precincts and polling places as the correct number of polling locations, would mean there was less than one poll worker per polling place. Puerto Rico included party observers in their count of poll workers, contrary to the Election Day Survey instructions. We report state-level responses for these jurisdictions, but exclude all highly questionable jurisdictions within these states or territories in the following tabulations. In addition to these administrative practices, we note that jurisdictions vary on how they staff polling places on Election Day. Some require that poll workers be present the entire day while other states schedule poll workers by shifts. These latter jurisdictions tend to report higher numbers of poll workers per polling place than other jurisdictions. In jurisdictions that consolidate precincts into a single polling place, some managerial positions may be shared among the consolidated precincts. These jurisdictions tend to report fewer poll workers per precinct than other jurisdictions. We do not exclude jurisdictions that report using poll workers in shifts or consolidate precincts because the information necessary to identify and control for these jurisdictions was not systematically collected on the Election Day Survey. In all, much care should be taken in interpreting the responses to the Election Day Survey regarding poll workers. Definitions of what constitutes a poll worker and a polling place or precinct are not consistent across jurisdictions and a few outlier jurisdictions, such as those in Louisiana and Illinois, figure prominently in the observed relationships. With this in mind, we present our primary findings. Jurisdictions with higher levels of income and education reported higher average numbers of poll workers per polling place or precinct and reported lower rates of staffing problems per precinct. Staffing problems appeared to be particularly acute for jurisdictions in the lowest income and education categories. Small, rural jurisdictions and large, urban jurisdictions tended to report higher rates of inadequate poll workers within polling places or precincts. Predominantly non-hispanic Black jurisdictions reported a greater percentage of polling places or precincts with inadequate number of poll workers. Predominantly non-hispanic Native American jurisdictions reported the second highest percentage of staffing problems. This appears to be related to similar reports on inadequate numbers of poll workers for Section 5 covered jurisdictions, though at least some of the observed relationships are attributable to the high percentage of understaffed polling places in Louisiana. Jurisdictions that anticipated Election Day needs reported higher average numbers of staffing of polling places or precincts and fewer instances of not being able to adequately staff polling places or

214 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-6 September 27, 2005 precincts. For example, jurisdictions in battleground states reported fewer polling places and precincts with inadequate staffing, as did jurisdictions that allow Election Day registration. Jurisdictions with no excuse absentee balloting and those with early voting reported lower rates of problems staffing polling places or precincts, perhaps because these alternative modes of voting reduced the Election Day burden for these jurisdictions. States Excluding Oregon, Washington reported the lowest number of poll workers per precinct, 1.5. However, Washington also consolidates many precincts and the high rate of absentee voting in the state reduces demands on polling places within the state. Washington also reported a middle-range average number 6.3 of poll workers per polling place. Washington reported that 7.3 percent of its polling places were inadequately staffed. The Virgin Islands and Oklahoma reported the lowest average number of poll workers per polling place, at 2.6 and 3.0, respectively. The Virgin Islands reported that 41.2 percent of its polling places were inadequately staffed, while Oklahoma reported no staffing problems. With 64.7 and 44.3 percent, respectively, of their polling places reported to have inadequate staffing, Louisiana and Hawaii reported the highest rates of staffing problems. Delaware and Illinois also reported a sizable percentage of polling places with inadequate staffing, 28.3 and 18.4 percent, respectively. In terms of absolute numbers, Illinois and Louisiana reported that over one thousand polling places or precincts had inadequate staffing: 1,693 and 1,550, respectively. Similar patterns among states exist when precincts are the unit of analysis. Even though Maryland reported 13.4 poll workers per polling place, the state also reported that 7.9 percent of polling places were inadequately staffed. We note that Maryland operates shifts of poll workers, so we do not know if the reported problems are for the entire day or specific shifts. Regions The U.S. Territories reported the lowest number of poll workers per polling place, 2.6. Within the United States, the West reported the lowest average number of poll workers per polling place, 6.6, with the Midwest, 6.7, and South, 7.8, reporting slightly higher numbers. The Northeast reported the highest average number of poll workers per polling place, In terms of average number of poll workers per precinct, the U.S. Territories reported the highest average, 14.5; followed by the Northeast, 9.1; the South, 6.6; the Midwest, 4.7; and the West, 4.1. The South reported the highest rate of inadequate staffing of polling places, at 8.1 percent, followed by the Midwest at 6.8 percent and the Northeast at 5.3 percent. The West reported the lowest rate of inadequately staffed polling places, at 2.3 percent. When examined from the perspective of precincts, the percentages are smaller and the regions retain their relative order. Urban to Rural The average number of poll workers per polling place was reported as 9.3 for urban and 7.4 for suburban jurisdictions, while small towns and rural areas reported lower averages, 6.7 and 5.3, respectively. The pattern is similar when precincts are the unit of analysis.

215 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-7 September 27, 2005 Urban jurisdictions also report the highest percentage of inadequate numbers of poll workers, 7.3 percent, followed by rural jurisdictions at 6.3 percent, small towns at 5.5 percent, and suburban jurisdictions at 3.6 percent. When then unit of analysis is precincts, the relative order is essentially the same. Size of Jurisdiction The reported average number of poll workers per polling place generally increases with jurisdiction size, from 4.8 for the smallest to 9.1 for the second largest, and dropping slightly to 7.7 for the largest jurisdictions. When the unit of analysis is the precinct, the same general pattern is evident. The percentage of jurisdictions reporting polling places with inadequate numbers of poll workers does not follow a clear pattern. Those jurisdictions with the smallest and largest populations report the largest percentages of polling places with inadequate numbers of poll workers, 9.2 and 7.4, respectively. Jurisdictions with voting age population (VAP) between 10,000 and 50,000 reported the next highest percentage, 6.8. Those in the 1,000-to-3,500 range reported the lowest percentage, 3.0 percent. The pattern is similar when precincts are the unit of analysis. Race and Ethnicity Predominantly non-hispanic Native American jurisdictions reported the lowest average number of poll workers per polling place, 5.7, and precinct, 5.5. Predominantly non-hispanic Black jurisdictions reported the highest average of poll workers per polling place, 9.6, and per precinct, 7.6. The high average number of poll workers per precinct for predominantly non-hispanic Black jurisdictions did not translate into better coverage of the polling places. Predominantly non-hispanic Black jurisdictions reported the highest percentage of inadequate numbers of staff, at 16.9 percent per polling place, and 12.8 percent per precinct. Predominantly non-hispanic Native American jurisdictions reported the second highest rate of inadequate staffing as a percentage of polling places, at 6.3 percent, and 6.2 percent per precinct. White jurisdictions reported percentage of inadequate staffing at 6.0 percent per polling place and 4.1 percent per precinct. Predominantly Hispanic jurisdictions reported the lowest rate of inadequate staffing, at 1.5 percent for polling places and 1.0 percent for precincts. Median Income The reported average number of poll workers per polling place tends to increase with median income of the jurisdiction, with 4.9 reported for the lowest category and 9.1 reported for the highest. There is a less apparent trend when the unit of analysis is the precinct, with 4.4 reported for the lowest category and 6.1 reported for the highest. In both cases, jurisdictions in the $35,000 $40,000 median income category report high averages, 8.2 and 5.7 percent respectively, for polling places and precincts, confounding the direct linear trend. The reported percentage of polling places with an inadequate number of poll workers generally follows a trend of decreasing percentages as median income within the jurisdictions rises. For the lowest income category, the very high rates of 23.5 percent of polling places and 16.1 percent of precincts reported inadequate numbers of poll workers. The numbers drop steeply as income rises, leveling off near 4 percent among polling places and 2.5 percent for precincts in jurisdictions with median income greater than $40,000.

216 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-8 September 27, 2005 High School Education The categories of reported average number of poll workers per polling place and precinct rise with education, from the lowest category reporting 4.8 and 4.1, respectively, steadily climbing to 7.8 and 5.6, respectively, for the second highest category, before falling slightly at the highest level of education, 7.6 or 4.7, respectively. Jurisdictions reporting inadequate numbers of poll workers are highest for the lowest education category, 20.8 percent among polling places and 14.2 percent among precincts. They generally follow a decreasing trend found across all jurisdictions as education rises, to 2.5 percent for polling places and 1.5 percent for precincts among jurisdictions in the highest education category. Section 203 Language Minority Requirements Jurisdictions covered by Section 203 reported a similar average number of poll workers per polling place as other jurisdictions, 7.9 and 7.2 respectively; for precincts, 5.4 and 5.3. Jurisdictions covered by Section 203 reported a similar percentage of polling places and precincts with an inadequate number of poll workers, 5.7 and 5.9 percent, respectively. Section 5 Preclearance of Voting Procedures Jurisdictions covered by Section 5 reported the same average number of poll workers per polling place, 7.4, as jurisdictions not covered by Section 5, and a slightly higher average number of poll workers per precinct than noncovered jurisdictions, 6.1 versus 5.2, respectively. Jurisdictions covered by Section 5 reported more than twice as high a percentage of polling places and precincts with an inadequate number of poll workers than jurisdictions not covered by Section 5: polling places scored 10.4 versus 4.5 percent, and precincts, 8.3 versus 3.1 percent. The relationship is primarily due to Louisiana s high percentage of inadequate poll workers per polling place or precinct. Type of Voting Equipment The reported average number of poll workers per polling place does not vary greatly by type of voting equipment, ranging from an average of 6.3 among optical scan jurisdictions to 9.8 among electronic jurisdictions. The range and order is similar when the unit of analysis is precincts: an average of 4.6 poll workers per precinct is reported for lever jurisdictions and a 7.2 average is reported for electronic jurisdictions. Among polling places, lever jurisdictions reported the highest percentage of polling places without an adequate number of poll workers, 36.0 percent. We note that the outlier state of Louisiana primarily uses lever machines. Punch card jurisdictions reported 10.6 percent of polling places with an inadequate number of poll workers, followed by 6.8 percent for electronic jurisdictions. Paper jurisdictions reported the lowest percentage, 1.8. The order is generally the same when precincts are the unit of analysis. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment reported a higher average number of poll workers per polling place, 8.1 versus 7.1 percent, and precinct, 6.1 versus 5.1 percent, than jurisdictions that did not change voting systems. Jurisdictions that changed voting equipment reported a slightly lower

217 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page 12-9 September 27, 2005 percentage of polling places, 4.0 versus 6.5, or precincts, 2.9 versus 4.5, without an adequate number of poll workers. Statewide Voter Registration Database Jurisdictions with a statewide voter registration database reported a lower average number of poll workers per polling place than other jurisdictions, 6.8 versus 7.6, and a slightly higher average number of poll workers per precinct than other jurisdictions, 5.6 and 5.3, respectively. Jurisdictions with a statewide voter registration database reported a much higher percentage of polling places or precincts without an adequate number of poll workers than other jurisdictions, 15.1 and 4.2 percent, respectively. We note that Louisiana has a statewide voter registration database. Election Day Registration Jurisdictions with Election Day registration reported a slightly lower average number of poll workers per polling place than other jurisdictions, 7.9 versus 7.4, and a higher average number of poll workers per precinct than other jurisdictions, 6.0 versus 5.3. Jurisdictions with Election Day registration reported a lower percentage than other jurisdictions of polling places, 2.1 versus 5.9 percent, and precincts, 1.6 versus 4.1 percent, without an adequate number of poll workers. However, caution should be taken in making any inferences because as we note, only 67 jurisdictions with Election Day registration reported the numbers required to calculate adequate number of poll workers and number of polling places or precincts. Provisional Ballot Acceptance Jurisdictions with precinct-only acceptance reported a higher average number of poll workers per polling place than other jurisdictions, 7.8 versus 7.0, and a higher average number of poll workers per precinct to within-jurisdiction acceptance, 5.7 versus 4.9 (those without provisional ballots reported the lowest numbers per polling place, 6.3, and the highest average numbers per precinct, 7.5). Jurisdictions that accept provisional ballots jurisdictionwide reported a higher percentage of polling places and precincts without an adequate number of poll workers, 9.9 and 6.6 percent, respectively, than jurisdictions that accept ballots cast only within polling places and precincts, 2.4 and 1.7 percent, respectively. No Excuse Absentee Balloting Jurisdictions with no excuse absentee balloting reported a slightly lower average number of poll workers per polling place and precinct than other jurisdictions, 7.0 versus 7.8 in polling places, and 4.8 versus 5.9 in precincts. Jurisdictions with no excuse absentee balloting reported a slightly lower percentage than other jurisdictions of inadequate numbers of staff in polling places, 5.2 versus 6.4 percent, and precincts, 3.5 versus 4.6 percent. Early Voting Jurisdictions with early voting reported a slightly lower average of poll workers per polling place than other jurisdictions, 7.1 versus 7.8, and a slightly higher per precinct, 5.4 versus 5.3. Compared with other jurisdictions, jurisdictions with early voting reported a markedly lower percentage of polling places 2.0 versus 10.4 percent and precincts 1.5 versus 6.7 percent without an adequate number of poll workers.

218 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page September 27, 2005 Battleground States Jurisdictions within battleground states reported a slightly higher average number of poll workers per polling place, 7.6 versus 7.3, than other jurisdictions and a slightly lower average of poll workers per precinct than other jurisdictions, 4.9 versus 5.7. Compared with other jurisdictions, jurisdictions within battleground states reported a lower percentage of polling places, 2.6 versus 7.4 percent, and precincts, 1.6 versus 5.5 percent, without an adequate number of poll workers. Presidential Margin of Victory According to presidential margin of victory, those jurisdictions in the second closest margin-ofvictory category reported the highest average number of poll workers per polling place and precinct, 13.3 and 8.7, respectively. All other jurisdictions reported averages around 7 percent per polling place and 5 percent per precinct. Jurisdictions in the second closest margin-of-victory category also reported the highest percentage of polling places without an adequate number of poll workers, 7.4 percent, followed by the jurisdictions with the closest margin of victory, at 7.0 percent. When measured in terms of precinct, the order is reversed, with the closest margin-of-victory jurisdictions reporting 4.9 percent of polling places with inadequate numbers of poll workers, and the second closest margin-of-victory category reporting 4.7 percent per precinct. All other jurisdictions reported a similar per-precinct average, except for jurisdictions in the third closest margin of victory category, which reported the lowest rates of inadequate numbers of poll workers, 1.9 and 1.4 percent for polling places and precincts, respectively. Red versus Blue Jurisdictions Jurisdictions won by Kerry by a majority, and those won by Bush by a plurality, reported higher average numbers of poll workers per polling place (10.8 where Kerry won 50 to 55 percent of the vote, and 8.5 where Kerry won 55 percent of the vote and above). For poll workers per precinct, those jurisdictions won by Bush by a plurality reported the highest number of poll workers per precinct: 7.5. Jurisdictions won by Bush by a plurality reported the highest percentage of polling places and precincts without an adequate number of poll workers, at 18.5 and 14.6 percent, respectively. All other jurisdictions reported a similar percentage, slightly below 5.1 percent for polling places and 4.2 for precincts. REFERENCES Formisano, Ronald P The Party Period Revisited. The Journal of American History 86(1): Lapomarda, Vincent A Maurice Joseph Tobin: The Decline of Bossism in Boston. The New England Quarterly 43(3): Oestreicher, Richard Urban Working-Class Political Behavior and Theories of American Electoral Politics, The Journal of American History 74(4): Sorauf, Frank J Extra-Legal Political Parties in Wisconsin. The American Political Science Review 48(3):

219 2004 Election Day Survey Report, Part 2 Survey Results Poll Workers, Page September 27, 2005 Willis, George L Electronic Vote Counting in a Metropolitan Area. Public Administration Review 26(1): Woodruff, Clinton Rogers A Review of Municipal Events, The American Journal of Sociology 13(4):

220 EAC Election Day Survey Poll Workers 2004 General Election Poll Workers Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :07:50 Total Total Number Percent Percent Election Total Number of Number of Average # of Average # of of Precincts or Precincts Polling Places Administration Number of Polling Poll Poll Workers Poll Workers Polling Places < Req Poll Cases < Req Poll Cases Code Name Jurisdictions Precincts Cases Places Cases Workers Cases Per Precinct Cases Polling Place Cases < Req Poll Workers Cases Workers Cases >100% Workers Cases >100% 01 Alabama 67 2, , , Alaska , Arizona 15 2, , , Arkansas 75 2, , , California 58 21, , , Colorado 64 3, , , Connecticut , Delaware , District of Columbia , Florida 67 6, , , Georgia 159 3, , , Hawaii , Idaho , Illinois , , , , Indiana 92 5, , , Iowa 99 1, , , Kansas 105 3, , , Kentucky 120 3, , , Louisiana 64 4, , , , Maine , Maryland 24 1, , , Massachusetts 351 2, , Michigan 83 5, , , Minnesota 87 4, Mississippi 82 1, , Missouri 116 5, , , Montana , Nebraska 93 1, , , Nevada 17 1, , New Hampshire New Jersey 21 6, , , New Mexico , New York 58 15, , North Carolina 100 2, , , North Dakota , Ohio 88 11, , , Oklahoma 77 2, , , Oregon 36 1, , Pennsylvania 67 24, Rhode Island , South Carolina 46 2, , South Dakota Tennessee 95 2, , , Texas 254 8, , , Utah 29 1, , , Vermont Virginia 134 2, , , Washington 39 6, , , West Virginia 55 1, , Wisconsin 1,910 3,563 1,253 2,686 1,596 18,669 1, , , Wyoming , American Samoa 1 66 Guam 1 72 Puerto Rico 110 1, , , Virgin Islands Total 6, ,252 5, ,754 5, ,962 4, , ,005 5,252 2, , , Maximum 1,910 21,857 1,253 14,467 1,596 99,289 1, , ,258 1, Average 119 3, , , Minimum Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

221 EAC Election Day Survey Poll Workers 2004 General Election Poll Workers Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :07:50 Total Total Number Percent Percent Election Total Number of Number of Average # of Average # of of Precincts or Precincts Polling Places Administration Number of Polling Poll Poll Workers Poll Workers Polling Places < Req Poll Cases < Req Poll Cases Code Name Jurisdictions Precincts Cases Places Cases Workers Cases Per Precinct Cases Polling Place Cases < Req Poll Workers Cases Workers Cases >100% Workers Cases >100% Election Administration Voting Equipment Used in 2004 General Election Poll workers removed from OR and PR because of questionable numbers and from ME, MI, TX, UT, VT, and WA where 0 Poll Workers were reported. None / Unknown , , , Punch card , , , , Lever , , , , Paper 1,734 5,704 1,573 3,416 1,183 26,116 1, , Optical scan 2,541 69,370 2,405 46,265 2, ,965 1, , ,690 1,178 1, , ,071 1 Electronic , , , , Multiple Systems , , , Changed Voting Equipment Since 2000 General Election Yes 1,753 46,241 1,296 31,649 1, ,173 1, , ,046 1, No 4, ,011 4,100 82,105 3, ,362 3, , ,813 4,211 1, , , State Wide Voter Registration System in Place Yes 1,335 33,575 1,153 20,815 1, , , No 5, ,677 4,243 92,939 4, ,023 3, , ,143 3,260 1, , ,842 1 Election Day Registration Yes 2,823 9,704 1,924 3,794 1,663 33,676 1, , , No 3, ,548 3, ,960 3, ,859 2, , ,534 5,229 2, , , Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,986 1,080 44,212 1, , , In Precinct Only 4, ,295 3,504 66,513 3, ,627 3, , ,975 1,082 1, , ,396 1 None 1,056 7, , , No Excuse Absentee Balloting Yes 3,781 70,535 3,106 47,225 2, ,571 2, , ,269 2,315 1, , , No 2, ,717 2,290 66,529 2, ,964 1, , ,590 2,937 1, , ,091 1 Early Voting Allowed Yes 1,701 69,882 1,683 51,609 1, ,926 1, , , , , ,199 1 No 4, ,370 3,713 62,145 3, ,609 3, , ,489 4,281 1, Covered By Section 203, Language Minority Requirements Yes , , , , No 6, ,201 4,953 77,656 4, ,755 4, , ,452 3,332 1, , , Covered By Section 5 of Voting Rights Act Yes , , , , No 5, ,276 4,541 88,074 4, ,277 3, , ,126 3,278 1, , ,671 2 Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

222 EAC Election Day Survey Poll Workers 2004 General Election Poll Workers Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :07:50 Total Total Number Percent Percent Election Total Number of Number of Average # of Average # of of Precincts or Precincts Polling Places Administration Number of Polling Poll Poll Workers Poll Workers Polling Places < Req Poll Cases < Req Poll Cases Code Name Jurisdictions Precincts Cases Places Cases Workers Cases Per Precinct Cases Polling Place Cases < Req Poll Workers Cases Workers Cases >100% Workers Cases >100% Demographics Region Northeast 1,710 25,068 1,230 13, , South 1,423 48,810 1,408 37,805 1, ,504 1, , ,234 2, Midwest 2,902 55,993 2,243 35,954 2, ,353 2, , ,038 1, West , , , Territories 113 1, , Urban to Rural Urban , , , , Suburban , , , Small Towns 1,710 41,994 1,421 28,085 1, ,288 1, , , Rural 3,307 32,252 2,780 21,938 2, ,245 2, , ,022 1,118 1, , , Not Available - Territories 113 1, , Size of Jurisdiction (VAP) < 1,000 1,761 2,118 1,229 1,350 1,169 6,579 1, , >=1,000 to <3,500 1,165 2, , , >=3,500 to <10,000 1,043 8, , , >=10,000 to <50,000 1,704 35,443 1,554 25,830 1, ,541 1, , ,251 1, >=50,000 to <250, , , , , >=250,000 to <1,000, , , , , >=1,000, , , , , Not Available 144 1, , Race and Ethnicity Predominently NH White 6, ,698 5, ,108 4, ,654 4, , ,749 4,905 2, , , Predominently NH Black 85 2, , , Predominently NH Native American , Predominently Hispanic 50 7, , , Not Available 145 1, , Median Income < $25, , , , >=$25,000 to <$30, , , , >=$30,000 to <$35,000 1,372 23,424 1,197 16,639 1, ,519 1, , >=$35,000 to <$40,000 1,215 40,250 1,056 24, , , >=$40,000 to <$45, , , , >=$45,000 to <$50, , , , >=$50,000 1,180 36, , , Not Available 151 1, , High School Education < 60% 126 2, , , >=60% to <70% , , , >=70% to <80% 1,646 51,393 1,411 32,782 1, ,590 1, , ,091 2, >=80% to <90% 3,111 87,644 2,502 56,581 2, ,176 2, , ,781 1, >=90% , , , Not Available 151 1, , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

223 EAC Election Day Survey Poll Workers 2004 General Election Poll Workers Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :07:50 Total Total Number Percent Percent Election Total Number of Number of Average # of Average # of of Precincts or Precincts Polling Places Administration Number of Polling Poll Poll Workers Poll Workers Polling Places < Req Poll Cases < Req Poll Cases Code Name Jurisdictions Precincts Cases Places Cases Workers Cases Per Precinct Cases Polling Place Cases < Req Poll Workers Cases Workers Cases >100% Workers Cases >100% Political Battleground States in 2004 Presidential Election Yes 3,093 59,123 2,113 33,037 2, ,662 2, , , No 3, ,129 3,283 80,717 2, ,873 2, , ,927 4,324 1, , , Margin of Victory in 2004 Presidential Election < 2.5% , , , >=2.5% to < 5.0% , , , >=5.0% to < 7.5% , , , >=7.5% to < 10.0 % 429 9, , , >=10.0 % 4, ,787 3,788 83,067 3, ,008 3, , ,855 4,016 1, , , Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,994 2,690 49,173 2, ,045 2, , ,277 1,880 1, , , Bush 50% to 55% , , , Bush < 50% 136 1, , , Kerry < 50% 150 4, , , Kerry 50% to 55% , , , Kerry > 55% 1,161 49, , , , Tied Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

224 Chapter 13 Polling Places Table 13 presents data from the Election Day Survey on precincts and polling places. The survey asked what constitutes a local election jurisdiction e.g., county, parish, township, or city and then asked for the number of local jurisdictions that provided information for the survey. The survey also asked for the number of precincts and polling places in each election jurisdiction. The term precincts in most states refers to the geographic area that covers a territory where voters would cast a unique ballot. Some states call these geographic areas voting districts, or wards, or beats. Precincts are usually built using registered voter records so that the precinct size stays under a limited size generally dictated by state law. The polling place is typically the specific building or location that voters go to each Election Day to cast their ballot. A polling place may serve several area precincts, but a single precinct usually only has a single polling place within it. In most areas of the country, there tends to be a one-to-one relationship between precincts and polling places, but this may not always be true. Applicability and Coverage Oregon conducts elections entirely by mail. However, provisions are made for voters to cast ballots at county election administrative offices, which is the number the state provided as its number of polling places. All other states had Election Day polling places for the 2004 election. The Election Day Survey unveiled some differences in how states treated precincts versus polling places. While most states reported data for both precincts and polling places, a handful of states reported data for only one item. For example, the state of Connecticut provided information only for polling places, not for precincts. On the other hand, the states of Georgia, Maine, Minnesota, South Carolina, and West Virginia only provided the number of precincts in each jurisdiction, but provided no information on the number of polling places. Historical Context Throughout United States history, voters have gathered at polling places on Election Day to determine the collective future course of the country. The method of voting has changed over time, from voters publicly stating their choice on county courthouse steps, to casting colorful party-printed ballots for all to see in ballot boxes, to the various methods of casting a ballot in secret. More recently, Oregon has done away with the polling place altogether, opting to run its federal elections entirely by mail, though the state still opens county administrative offices for people to vote inperson. Jurisdictions in other states also have begun reviewing the option of conducting their elections entirely by mail, but in most instances this has been for smaller, local elections. Following the 2004 elections, more jurisdictions in Washington State have moved to all mail elections. An issue regarding election administration of polling places is the efficient distribution of resources to ensure a fair and accurate election that provides satisfactory service to the voters. The issue is not a new one. For example, the expansion of women s suffrage prompted several states to provide

225 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-2 September 27, 2005 resources to increase the number of poll workers per polling place in anticipation of a greater volume of voters (West 1921). In 1968, the Office of the Mayor of New York conducted a pilot study to test the efficacy of drawing voting precincts by computer to reduce lines at the polls, equalize voting delays, and reduce the cost of conducting elections (Savas 1971). The study ultimately recommended that blocks within the city be split in order to equalize registration among voting precincts and thereby more efficiently distribute resources, which netted a savings to the city of $2 million (Savas, Lipton, and Burkholz 1972). This representative study was publicly published in an academic journal, and we are certain that states and localities have conducted similar internal studies of their election administration. Over the past two decades, Election Data Services has collected the number of precincts for each election. The nationwide numbers going back to 1980 are in Table 13a. The number of precincts in the United States has gradually grown with the growth in population and registration. There has traditionally been a larger increase in the year immediately after redistricting takes place, as election administrators adjust precinct boundaries that need to be split apart due to new district boundaries. The year 2004 marked a significant drop in the overall number of precincts, possibly due to the higher costs of new voting equipment. The U.S. Election Assistance Commission (EAC) survey total of 174,252 precincts for 2004 is lower than it should be due to lack of data from the states of Connecticut, New Hampshire, and Pennsylvania, but it does include Puerto Rico, which is not in the Election Data Services dataset. Table 13a. Number of Precincts Nationwide, Number of Election Year Precincts , , , , , , , , , , ,037 A secondary source of the number of precincts in selected states is available through the national census of the population. Beginning with the 1980 census, the Bureau of the Census implemented a voluntary program (PL ) whereby states could obtain population counts for geographic areas that roughly approximated precincts. This allows states to align their voting precinct boundaries with census geography to facilitate the merging of census and election data for redistricting purposes. There were rough approximations because the bureau guidelines dictated that the states had to use whole census blocks to build what they called the voting tabulation districts (or VTDs). Some states merged precincts together to form mega VTDs as a way of getting around the whole block requirement. These VTDs roughly approximated the precincts used in the general election two years before the census (i.e., the 1978 elections, the 1988 elections, and the 1998 elections). VTDs are not

226 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-3 September 27, 2005 updated following the election or redistricting, nor are they maintained by the Census Bureau. In advance of the 2000 census, all states except California, Florida, Kentucky, Montana, North Dakota, Ohio, Oregon, and Wisconsin participated in this program, and among the territories, only Puerto Rico participated. Within the United States, a total of 127,605 VTDs were reported to the Bureau of the Census in preparation for the 2000 census. Puerto Rico reported 1,714 VTDs. The number of voting precincts is not static. Population and registration changes often necessitate the splitting or merging of existing precincts. After a redistricting, precincts that are split by a new district boundary often need to be reconfigured to ensure the uniformity of the ballot throughout the precinct. However, some states do not change their precinct boundaries following redistricting, and instead have what they call split precincts, which are divided by some upper level of political or legal geography (i.e., state legislative boundaries, city boundaries, etc.). Poll workers in split precincts must correctly identify which part of the precinct a voter resides within, so that they are provided their correct ballot configuration. Voters receiving incorrect ballots are among the problems reported in the 2004 election. Survey Results Table 13 presents data on precincts and polling places from questions 19 and 20 on the Election Day Survey. In the table, the average numbers of precincts per polling place and polling places per precinct are calculated as well as the average total registration and voting age population (VAP) per precinct and polling place. The column headings in Table 13 are as follows:

227 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-4 September 27, 2005 Column Headings for Table 13. Polling Places Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Number of Precincts Number of precincts from survey question 19 5 Cases Number of jurisdictions that responded to question 19 6 Total Number of Polling Places Number of polling places from survey question 20 7 Cases Number of jurisdictions that responded to question 20 8 Average # of Precincts in a Polling Place 9 Average # of Polling Places in a Precinct Number of precincts (col. 4) divided by the number of polling places (col. 6) Number of polling places (col. 6) divided by the number of precincts (col. 4) 10 Cases Number of jurisdictions that responded to questions 19 and Total Registration Number of active and inactive registered voters, number of persons who voted on Election Day in six states, and VAP data for North Dakota and jurisdictions in Wisconsin that do not have voter registration, from col. 4 of table 2 12 Cases Number of jurisdictions that responded to survey question 1, provided Election Day registration data, or for which VAP data was substituted for voter registration data 13 Average Registration per Precinct Number of registered voters (col. 11) divided by the number of precincts (col. 4) 14 Cases Number of jurisdictions that responded to survey questions 1 and 19, provided Election Day registration data, or for which VAP data was substituted for voter registration data

228 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-5 September 27, 2005 Column Headings for Table 13 (cont.) Col. Heading Description 15 Voting Age Population Estimated November 2004 VAP 16 Cases Number of jurisdictions for which 2004 VAP was constructed 17 Average Voting Age Population per Precinct Estimated VAP (col. 15) divided by the number of precincts (col. 4) 18 Cases Number of jurisdictions for which November 2004 VAP estimates were compiled and that responded to question Average Registration per Polling Place Number of registered voters (col. 11) divided by the number of polling places (col. 6) 20 Cases Number of jurisdictions that responded to survey questions 1 and 20, provided Election Day registration data, or for which VAP data was substituted for voter registration data 21 Average Voting Age Population per Polling Place Estimated VAP (col. 15) divided by the number of polling places (col. 6) 22 Cases Number of jurisdictions for which November 2004 VAP estimates were compiled and that responded to question 6 Analysis of Survey Results The following is our analysis of the data in Table 13 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 13 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation.

229 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-6 September 27, 2005 Summary The number of voting precincts and the number of polling places are often not the same within a jurisdiction. There are several reasons for this. In a number of jurisdictions, the county courthouse is also designated as a polling place. In addition, some jurisdictions added in their early voting sites as additional polling places in their reported data. Finally, some jurisdictions said they just had polling places and not precincts and some appeared to be confused by the terminology that was foreign to their state. In all, 383 jurisdictions reported a number of polling places larger than their number of precincts. Sometimes, two or more voting precincts will be consolidated, or share the same polling place. Jurisdictions reporting more precincts than polling places totaled 1,576, from which we might infer that at least this many consolidated polling places existed in the 2004 elections. There are two ways to express the ratio of polling places and precincts, with either number used in the numerator or the denominator. Here, we discuss the ratio of the average number of precincts in a polling place and provide the other ratio for completeness. There is evidence that the ratio of precincts to polling places is related to the urban and rural character of the state, the socioeconomic characteristics of the jurisdiction, and the factors related to the Election Day experience, such as Election Day registration. In urban areas precinct consolidation is easier, and perhaps necessary, due to limited availability of suitable locations for polling places in dense population areas. We find higher reported ratios of precincts to polling places in urban areas, and by a consequence states and regions with larger urban populations. Other tabulations associated with urban/rural character, such as vote for presidential winner, report similar relationships. Income and education of a jurisdiction are also related, with higher reported ratios of precincts to polling places at higher levels of education and income. This is not simply a consequence of the urban/rural character of the jurisdiction. For some states, pressures are relieved in Election Day polling places through other methods of voting. Oregon, which conducts its election entirely by mail, needs one polling place per county. States with Election Day registration also consolidate fewer precincts than those without, perhaps to aid in the processing of voters at the polls on Election Day. The best determinant of the distribution of polling places among voters is to divide the number of registered voters that are serviced by the number of voting precincts and polling places that service them. The polling places per registration will be the primary measure used in this analysis, although additional measures for precincts and dividing both precincts and polling places by VAP are provided. Excluding Oregon, the strongest reported relationship between average registration per polling place is found in the population size of the jurisdiction. Jurisdictions of smaller size report a smaller number of registered voters per polling place. This size of the jurisdiction is related to the observed relationships explored in other tabulations, such as the urban/rural character of the jurisdiction, the region the jurisdiction is located in, the type of equipment used, and the presidential winner of the jurisdiction.

230 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-7 September 27, 2005 There is also a relationship between income and education, with lower reported average registration per jurisdiction for lower levels of income and education. Here, it is useful to compare the jurisdiction s average registration per polling place with the average citizen voting age population (CVAP) per polling place, since persons of lower income and education tend to participate at lower rates. This is partially responsible for the relationship between income and education since, for example, among education categories the reported ratio of the average registration to the average CVAP per polling place is 73 percent for the lowest education category and 90 percent for the highest category. But this is not a complete explanation of income and education disparities since average CVAP per polling place rises with education categories, like registration (but not with the same rate of increase). Finally, there is a relationship between service demands in polling places and average registration per polling place, as those jurisdictions with Election Day registration have lower registration per polling place than other jurisdictions and those with early voting report higher average registration per polling place. States Excluding Oregon, Washington reported the highest ratio of voting precincts to polling places, 4.18:1, but this may correspond to the state having the highest rate of absentee ballots used. [See chapter 5.] Nevada reported a similar high ratio, 3.01:1, but it also reported a large proportion of early voting. In these states, fewer demands are placed on polling places on Election Day. New York also reported a high ratio, 2.25:1, due primarily to consolidation of precincts in and around New York City. Most other states reported an average ratio of between one and two voting precincts per polling place, with nearly all clustered around one. Excluding Oregon, states generally range from slightly under 1,000 to slightly over 2,000 reported registered voters per polling place. Massachusetts reported the highest average, 2,811, and Wyoming reported the smallest average, 794. Regions The ratio of precincts to polling palaces by region is strongly affected by the states with high ratios mentioned above. The Northeast and West reported the highest ratios, 1.97:1 and 1.65:1, respectively, and the Midwest and South reported the smallest ratios, 1.43:1 and 1.18:1, respectively. The average registration per polling place among regions is highest for the Northeast at 1,747 and lowest for the Midwest at 1,125. Urban to Rural The distribution of the ratio of polling palaces to precincts is related to the size of the jurisdiction. There is a near linear decrease in the reported ratio of precincts to polling places from urban to rural jurisdictions, from 1.62 for urban to 1.30 for rural jurisdictions. The average registration per polling place is also strongly related to the size of the jurisdiction. Rural areas reported almost half of the average registration per polling place than urban and suburban jurisdictions, 809 versus 1,587. Size of Jurisdiction Similar to the urban/rural tabulation, there is a near linear decrease in the reported ratio of precincts to polling places from larger to smaller jurisdictions, ranging from 1.82:1 for the largest population

231 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-8 September 27, 2005 jurisdictions to 1.25:1 for the second smallest. The smallest jurisdictions deviate from the trend, with a reported 1.62 precincts per polling place. The reported average registration per polling place is even more strongly related to the population size of the jurisdiction than to the urban/rural character. Jurisdictions with less than 1,000 VAP reported an average registration per polling place of 461, while those with 50,000 and greater reported an average registration per polling place slightly higher than 1,500. Race and Ethnicity Among racial and ethnicity categories, the reported ratio of precincts to polling places is highest for predominantly Non-Hispanic White jurisdictions, 1.47:1. Predominantly Hispanic jurisdictions reported the next highest ratio, 1.35:1, followed by predominantly Non-Hispanic Black, 1.20:1, and predominantly Non-Hispanic Native American, 1.03:1. This relationship remains among racial and ethnicity categories when the states of Nevada, Oregon, and Washington are removed from the tabulation. Among race and ethnicity categories, all but predominantly Non-Hispanic Native American jurisdictions reported an average registration per polling place slightly above 1,300. Predominantly Non-Hispanic Native American jurisdictions reported an average of 749, which may reflect the rural character of reservations. Median Income Among income categories, the reported ratio of precincts to polling places for jurisdictions rises as income increases, from 1.15:1 for the lowest income category of under $25,000 until reaching $35,000, where the ratio remains relatively constant around 1.5:1. Among income categories, the reported average registration per polling place increases from 692 for the lowest category to around 1,500 at $35,000 and above. High School Education Among education categories, the reported ratio of precincts to polling places for jurisdictions rises nearly linearly as education rises, from 1.11:1 to 1.66:1. Among education categories, the reported average registration per polling place increases from 915 for the lowest education category to 1,771 for the highest category. Section 203 Language Minority Requirements In comparing Section 203 covered jurisdictions with other jurisdictions, the ratio of precincts to polling places is similar, 1.48:1 and 1.43:1, respectively. Registration per polling place is nearly equal, too, at 1,348 and 1,408, respectively. Section 5 Pre-clearance of Voting Procedures Among Section 5 covered jurisdictions, the ratio of precincts to polling places is lower than among noncovered jurisdictions, 1.19:1 and 1.52:1, respectively. Among Section 5 covered jurisdictions, the average registration per polling place is higher than other jurisdictions, 1,483 and 1,361, respectively.

232 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page 13-9 September 27, 2005 Type of Voting Equipment Among categories of voting equipment, the reported ratio of precincts to polling places is fairly equal at 1.4:1 across jurisdictions, except for those that use lever machines, which reported a ratio of 2.01:1. This relationship is primarily driven by the use of lever machines in New York City, which has a high number of consolidated precincts. Among categories of voting equipment, the reported average registration per polling place is lowest for jurisdictions that use paper, at 671. This is primarily a consequence of the higher usage of paper ballots in smaller population jurisdictions. The next lowest is punch card jurisdictions at 1,094. The highest average is reported by jurisdictions that use multiple systems, at 1,936, followed by lever, at 1,549, and electronic machine jurisdictions, at 1,470. Changed Voting Equipment since 2000 Jurisdictions that changed voting equipment reported a slightly lower ratio of precincts to polling places than other jurisdictions, 1.37:1 and 1.48:1, respectively. Among those jurisdictions that changed voting equipment, the reported average registration per polling place is higher than those that did not, 1,475 and 1,355, respectively. Statewide Voter Registration Database Among jurisdictions with a statewide voter registration database, the reported ratio of precincts to polling places is lower than those jurisdictions without, 1.26:1 and 1.49:1, respectively. Among those jurisdictions with a statewide voter registration database, the reported average registration per polling place is slightly higher than those without, 1,485 and 1,367, respectively. Election Day Registration Among jurisdictions with Election Day registration, the reported ratio of precincts to polling places is very similar to those jurisdictions without, 1.49:1 and 1.45:1, respectively. Among those jurisdictions with Election Day registration, the reported average registration per polling place is also very similar to those that do not register on Election Day, at 1,355 and 1,389, respectively. Provisional Ballot Acceptance Among jurisdictions according to the method of accepting provisional ballots, the reported ratio of precincts to polling places is similar across jurisdictions for which provisional ballots are accepted jurisdiction-wide and in-precinct, 1.46:1 and 1.45:1, respectively. Those that do not have provisional ballots, which tend to be those with Election Day registration, reported a lower ratio, 1.08:1. Among jurisdictions according to the method of accepting provisional ballots, the reported average registration per polling place is notably lower in jurisdictions for which provisional ballots are accepted jurisdiction-wide vs. in-precinct jurisdictions, 1,274 and 1,468, respectively. This would seem to indicate that communities that accept provisional ballots jurisdiction-wide purposely keep their precinct sizes low to accommodate the potential of other voters showing up to vote. Those that do not have provisional ballots, which tend to be those with Election Day registration, reported a lower average registration per polling place: 1,286.

233 2004 Election Day Survey Report, Part 2 Survey Results Polling Places, Page September 27, 2005 No Excuse Absentee Balloting Among jurisdictions with no excuse absentee balloting, the reported ratio of precincts to polling places is similar to those jurisdictions without, 1.47:1 and 1.43:1, respectively. Among jurisdictions with no excuse absentee balloting, the reported average registration per polling place is lower than in those jurisdictions without, at 1,318 and 1,438. Early Voting Among jurisdictions with early voting, the reported ratio of precincts to polling places is lower than those jurisdictions without, 1.30:1 and 1.57:1, respectively. Among jurisdictions with early voting, the reported average registration per polling place is very similar to other jurisdictions, 1,384 and 1,392, respectively. Battleground States Among jurisdictions in battleground states, the reported ratio of precincts to polling places is higher than other nonbattleground jurisdictions, 1.62:1 and 1.38:1, respectively. Among jurisdictions in battleground states, the reported average registration per polling place is higher than those jurisdictions not in battleground states, 1,525 and 1,332, respectively. Presidential Margin of Victory Among jurisdictions tabulated by presidential margin of victory, the reported ratio of precincts to polling places follows no clear pattern, varying between 1.43:1 and 1.58:1. Among jurisdictions tabulated by presidential margin of victory, there is a clear linear pattern: the reported average registration per polling place is highest for the closest margin, 1,650, then declines to 1,546 for second closest margin, and is between 1,390 and 1,344 for the remaining jurisdictions. Red versus Blue Jurisdictions Among jurisdictions tabulated by presidential winner, the reported ratio of precincts to polling places is lower in jurisdictions won by Bush than by Kerry, ranging between 1.28:1 and 1.40:1 for jurisdictions won by Bush and 1.47:1 and 1.53:1 for Kerry. Part of the reason for this relationship is the concentration of Kerry supporters in urban areas where there is greater precinct consolidation. Among jurisdictions tabulated by presidential winner, the reported average registration per polling place is lower in jurisdictions won by Bush than by Kerry, ranging between 1,269 and 1,466 for jurisdictions won by Bush and 1,450 and 1,637 for Kerry. This is related to the concentration of Bush supporters in small population jurisdictions that tend to have lower registration per polling place. REFERENCES Savas, E. S A Computer-Based System for Forming Efficient Election Districts. Operations Research 19(1): Savas, E. S., Harry Lipton, Leonora Burkholz Implementation of an OR Approach for Forming Efficient Election Districts. Operations Research 20(1): West, Victor J Legislation Respecting Elections. The American Political Science Review 16(3):

234 EAC Election Day Survey Polling Places 2004 General Election Polling Places Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :08:14 Total Average # Average # Average Average Average Average Election Total Number of of Precincts of Polling Registration Voting Age Registration Voting Age Administration Number of Polling In A Polling Places In Total Per Voting Age Population per per Population per Code Name Jurisdictions Precincts Cases Places Cases Place A Precinct Cases Registration Cases Precinct Cases Population Cases Precinct Cases Polling Place Cases Polling Place Cases 01 Alabama 67 2, , ,597, , ,425, , , , Alaska , , , , , , Arizona 15 2, , ,642, , ,194, , , , Arkansas 75 2, , ,699, ,069, , California 58 21, , ,646, ,647, , , , Colorado 64 3, , ,101, ,456, , , , Connecticut ,831, ,684, , , Delaware , , , , , , District of Columbia , , , , , , Florida 67 6, , ,300, , ,441, , , , Georgia 159 3, , ,248, , ,534, , , , Hawaii , , , , , , Idaho , ,025, , , , Illinois , , ,195, ,518, , Indiana 92 5, , ,296, ,635, , , Iowa 99 1, , ,226, , ,274, , , , Kansas 105 3, , ,695, ,049, , Kentucky 120 3, , ,794, ,157, , Louisiana 64 4, , ,932, ,358, , , Maine ,026, , ,037, , Maryland 24 1, , ,105, , ,200, , , , Massachusetts 351 2, , ,098, , ,956, , , , Michigan 83 5, , ,164, , ,616, , , , Minnesota 87 4, ,977, ,872, Mississippi 82 1, , ,469, ,139, , , Missouri 116 5, , ,194, ,344, , , Montana , , , Nebraska 93 1, , ,160, ,316, Nevada 17 1, ,073, ,737, , , , New Hampshire , ,000, New Jersey 21 6, , ,011, ,573, , , , New Mexico , ,402, , New York 58 15, , ,837, ,790, , , North Carolina 100 2, , ,526, , ,414, , , , North Dakota , , Ohio 88 11, , ,965, ,680, , , Oklahoma 77 2, , ,143, ,664, , , , Oregon 36 1, ,141, , ,766, , , , Pennsylvania 67 8,366, ,615, Rhode Island , , , , , , South Carolina 46 2, ,318, , ,174, , South Dakota , , Tennessee 95 2, , ,748, , ,516, , , , Texas 254 8, , ,098, , ,263, , , , Utah 29 1, , ,278, ,645, , , Vermont , , , , , , Virginia 134 2, , ,515, , ,695, , , , Washington 39 6, , ,508, ,732, , , West Virginia 55 1, ,168, ,430, Wisconsin 1,910 3,563 1,253 2,686 1, ,247 4,179,774 1, ,252 4,188,206 1, ,252 1, ,584 1, , Wyoming , , , American Samoa 1 66 Guam 1 72 Puerto Rico 110 1, , ,440, , , Virgin Islands , , Total 6, ,252 5, ,754 5, , ,265,030 6, , ,279,989 6,425 1, ,273 1, ,160 1, ,160 Maximum 1,910 21,857 1,253 14,467 1, ,247 16,646,555 1,894 2, ,252 26,647,955 1,894 3, ,252 59, ,584 76, ,584 Average 119 3, , ,344, , ,338, , , , Minimum , , Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

235 EAC Election Day Survey Polling Places 2004 General Election Polling Places Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :08:14 Total Average # Average # Average Average Average Average Election Total Number of of Precincts of Polling Registration Voting Age Registration Voting Age Administration Number of Polling In A Polling Places In Total Per Voting Age Population per per Population per Code Name Jurisdictions Precincts Cases Places Cases Place A Precinct Cases Registration Cases Precinct Cases Population Cases Precinct Cases Polling Place Cases Polling Place Cases Election Administration Voting Equipment Used in 2004 General Election None / Unknown , , ,484, , ,612, , , , Punch card , , ,767, ,552, , , Lever , , ,662, ,918, , , Paper 1,734 5,704 1,573 3,416 1, ,169 3,085,167 1, ,572 3,308,339 1, , , ,182 Optical scan 2,541 69,370 2,405 46,265 2, ,179 69,198,628 2, ,399 88,323,954 2,541 1, ,405 1, ,179 1, ,179 Electronic , , ,068, , ,761, , , , Multiple Systems , , ,997, , ,803, , , , Changed Voting Equipment Since 2000 General Election Yes 1,753 46,241 1,296 31,649 1, ,093 51,149,755 1,746 1, ,293 69,121,688 1,747 1, ,296 1, ,265 2, ,265 No 4, ,011 4,100 82,105 3, , ,115,275 4, , ,158,301 4,678 1, ,977 1, ,895 1, ,895 State Wide Voter Registration System in Place Yes 1,335 33,575 1,153 20,815 1, ,384,852 1,321 1, ,152 48,152,870 1,335 1, ,153 1, ,132 1, ,132 No 5, ,677 4,243 92,939 4, , ,880,178 5, , ,127,119 5,090 1, ,120 1, ,028 1, ,028 Election Day Registration Yes 2,823 9,704 1,924 3,794 1, ,314 10,323,368 2, ,923 11,509,789 2, ,912 1, ,651 1, ,651 No 3, ,548 3, ,960 3, , ,941,662 3, , ,770,200 3,632 1, ,361 1, ,509 1, ,509 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,986 1,080 44,212 1, ,070 65,077,741 1, ,073 88,988,159 1,162 1, ,080 1, ,063 1, ,063 In Precinct Only 4, ,295 3,504 66,513 3, , ,336,604 4, , ,866,238 4,334 1, ,503 1, ,889 1, ,889 None 1,056 7, , ,850,685 1, ,425, , , , No Excuse Absentee Balloting Yes 3,781 70,535 3,106 47,225 2, ,572 64,333,790 3, ,104 85,693,320 3,754 1, ,094 1, ,909 1, ,909 No 2, ,717 2,290 66,529 2, , ,931,240 2, , ,586,669 2,671 1, ,179 1, ,251 1, ,251 Early Voting Allowed Yes 1,701 69,882 1,683 51,609 1, ,617 73,710,075 1,686 1, ,682 99,654,623 1,701 1, ,683 1, ,617 1, ,617 No 4, ,370 3,713 62,145 3, , ,554,955 4, , ,625,366 4,724 1, ,590 1, ,543 1, ,543 Covered By Section 203, Language Minority Requirements Yes , , ,756, ,670, , , , No 6, ,201 4,953 77,656 4, , ,508,534 6, , ,609,924 5,957 1, ,830 1, ,718 1, ,718 Covered By Section 5 of Voting Rights Act Yes , , ,868, , ,030, , , , No 5, ,276 4,541 88,074 4, , ,396,175 5, , ,249,505 5,546 1, ,418 1, ,358 1, ,358 Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

236 EAC Election Day Survey Polling Places 2004 General Election Polling Places Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :08:14 Total Average # Average # Average Average Average Average Election Total Number of of Precincts of Polling Registration Voting Age Registration Voting Age Administration Number of Polling In A Polling Places In Total Per Voting Age Population per per Population per Code Name Jurisdictions Precincts Cases Places Cases Place A Precinct Cases Registration Cases Precinct Cases Population Cases Precinct Cases Polling Place Cases Polling Place Cases Demographics Region Northeast 1,710 25,068 1,230 13, ,273,670 1, ,230 41,988,043 1,696 1, ,219 1, , South 1,423 48,810 1,408 37,805 1, ,302 62,606,676 1,407 1, ,407 79,567,761 1,423 1, ,408 1, ,301 1, ,301 Midwest 2,902 55,993 2,243 35,954 2, ,140 44,048,138 2, ,236 49,563,034 2, ,242 1, ,472 1, ,472 West , , ,845, ,161, , , , Territories 113 1, , ,490, , , Urban to Rural Urban , , ,441, ,075, , , , Suburban , , ,552, , ,268, , , , Small Towns 1,710 41,994 1,421 28,085 1, ,177 44,193,768 1, ,419 56,213,989 1,700 1, ,421 1, ,271 1, ,271 Rural 3,307 32,252 2,780 21,938 2, ,354 19,586,556 3, ,775 23,722,427 3, , , ,543 Not Available - Territories 113 1, , ,490, , , Size of Jurisdiction (VAP) < 1,000 1,761 2,118 1,229 1,350 1, ,006 1, , ,315 1, , , ,168 >=1,000 to <3,500 1,165 2, , ,182,148 1, ,267,899 1, >=3,500 to <10,000 1,043 8, , ,966,645 1, ,692,594 1, >=10,000 to <50,000 1,704 35,443 1,554 25,830 1, ,422 31,472,681 1, ,549 38,463,619 1, ,554 1, ,503 1, ,503 >=50,000 to <250, , , ,992, , ,558, , , , >=250,000 to <1,000, , , ,396, , ,995, , , , >=1,000, , , ,867, ,402, , , , Not Available 144 1, , ,492, , , Race and Ethnicity Predominently NH White 6, ,698 5, ,108 4, , ,662,585 6, , ,258,977 6,262 1, ,125 1, ,917 1, ,917 Predominently NH Black 85 2, , ,098, , ,061, , , , Predominently NH Native American , , Predominently Hispanic 50 7, , ,749, , ,658, , , , Not Available 145 1, , ,523, , , , Median Income < $25, , , ,504, ,079, >=$25,000 to <$30, , , ,917, ,220, , >=$30,000 to <$35,000 1,372 23,424 1,197 16,639 1, ,019 22,970,583 1, ,195 28,691,481 1,372 1, ,197 1, ,074 1, ,074 >=$35,000 to <$40,000 1,215 40,250 1,056 24, ,443,694 1, ,056 50,829,468 1,215 1, ,056 1, , >=$40,000 to <$45, , , ,780, ,717, , , , >=$45,000 to <$50, , , ,218, , ,092, , , , >=$50,000 1,180 36, , ,936,586 1,178 1, ,649,351 1,179 1, , , Not Available 151 1, , ,492, , , High School Education < 60% 126 2, , ,817, ,401, , , >=60% to <70% , , ,944, ,653, , , , >=70% to <80% 1,646 51,393 1,411 32,782 1, ,225 49,285,773 1, ,406 64,350,042 1,646 1, ,411 1, ,314 1, ,314 >=80% to <90% 3,111 87,644 2,502 56,581 2, ,124 93,198,279 3, , ,912,781 3,111 1, ,502 1, ,408 1, ,408 >=90% , , ,495, , ,930, , , , Not Available 151 1, , ,523, , , , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

237 EAC Election Day Survey Polling Places 2004 General Election Polling Places Cases = Number of Jurisdictions Reporting Subject Matter Updated: 09/19/ :08:14 Total Average # Average # Average Average Average Average Election Total Number of of Precincts of Polling Registration Voting Age Registration Voting Age Administration Number of Polling In A Polling Places In Total Per Voting Age Population per per Population per Code Name Jurisdictions Precincts Cases Places Cases Place A Precinct Cases Registration Cases Precinct Cases Population Cases Precinct Cases Polling Place Cases Polling Place Cases Political Battleground States in 2004 Presidential Election Yes 3,093 59,123 2,113 33,037 2, ,959 64,166,639 3, ,111 76,824,163 3,074 1, ,112 1, ,296 1, ,296 No 3, ,129 3,283 80,717 2, , ,098,391 3, , ,455,826 3,351 1, ,161 1, ,864 1, ,864 Margin of Victory in 2004 Presidential Election < 2.5% , , ,923, , ,185, , , , >=2.5% to < 5.0% , , ,133, ,781, , , , >=5.0% to < 7.5% , , ,830, ,701, , , , >=7.5% to < 10.0 % 429 9, , ,833, ,292, , , , >=10.0 % 4, ,787 3,788 83,067 3, , ,044,988 4, , ,310,466 4,486 1, ,785 1, ,658 1, ,658 Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,994 2,690 49,173 2, ,445 68,178,580 3, ,686 86,412,155 3,112 1, ,688 1, ,612 1, ,612 Bush 50% to 55% , , ,682, ,877, , , , Bush < 50% 136 1, , ,041, , ,380, , , , Kerry < 50% 150 4, , ,850, , ,883, , , , Kerry 50% to 55% , , ,160, ,466, , , , Kerry > 55% 1,161 49, , ,846,628 1, ,245,074 1,159 1, , , Tied , , Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

238 Chapter 14 Disability The U.S. Election Assistance Commission s (EAC) final area of inquiry on the Election Day Survey was disability issues surrounding precincts and polling places. Besides establishing base numbers of precincts and polling places, the EAC sought information on three different types of disability related to voting systems and polling locations. Question 21 on the survey asked for the number of polling places: (1) that could be accessed by a voter who uses a wheelchair, 2) where a visually impaired voter could cast a private ballot, and 3) where a physically disabled voter could cast a ballot on an accessible voting system. Applicability and Coverage As noted earlier in chapter 13 on polling places, the Election Day Survey unveiled some differences in how states treated precincts versus polling places. While most states reported data for both precincts and polling places, a handful of states reported data for only one item. For example, the state of Connecticut provided information only for polling places, not for precincts. On the other hand, the states of Georgia, Maine, Minnesota, South Carolina, and West Virginia only provided the number of precincts in each jurisdiction, but provided no information on the number of polling places. Because of the different ways in which states responded to question 21, we have calculated all data for our analysis for both precincts and polling places. The most significant issue in this chapter is the overall lack of data. Only 26 of the 55 states and territories provided information on disability in response to question 21. While a greater number of polling places were reported to be wheelchair-accessible (question 21a), the much smaller numbers of polling places reported to be available to the visually impaired (question 21b) or physically disabled (question 21c) may have resulted from how the survey questions were worded. Some states reported that they interpreted the last two questions as seeking information on the voting equipment in use and its accessibility, rather than the physical configuration of the polling place. Historical Context Federal laws on voting assistance to persons with physical and sensory disabilities extend back at least 40 years to the Voting Rights Act of Under Section 208 of the Act, voters requiring assistance to vote by reason of blindness, disability, or inability to read or write were entitled to assistance by a person of the voter s choice. The next major piece of legislation focused on the accessibility of polling places and voter registration facilities. The Voting Accessibility for the Elderly and Handicapped Act of 1985 (P.L ) required election jurisdictions to ensure that all polling places for federal elections were accessible to elderly and handicapped voters, and required jurisdictions to provide a reasonable number of accessible permanent registration facilities, unless alternatives such as registration by mail were available. States also were required to make registration and voting aids such as large-type instructions and information by telecommunications devices for the deaf (TDDs) available to disabled voters and to report the number of accessible and inaccessible polling places to the Federal Election Commission (FEC). Until then, states had

239 2004 Election Day Survey Report, Part 2 Survey Results Disability, Page 14-2 September 27, 2005 generally relied on procedures such as absentee voting to serve the elderly and persons with disabilities, although by 1984, 22 states had laws on polling place accessibility. Discrimination against individuals with disabilities in voting, as well as housing, education, employment, transportation, public accommodations, and other areas was mentioned in the preamble to the Americans with Disabilities Act (ADA) (P.L ), enacted in Regulations promulgated by the U.S. Department of Justice applied the ADA Standards for Accessible Design (28 CFR Part 36) to polling places. The ADA standards addressed issues such as parking, passenger drop-off areas, sidewalks and walkways, and building entrances and corridors for voters using wheelchairs or other mobility devices as well as voters who are blind or have low vision. One of the basic purposes of the National Voter Registration Act of 1993 (NVRA), also known as Motor Voter, was to increase historically low registration rates among minorities and persons with disabilities by requiring government offices providing services to persons with disabilities to provide all program applicants with voter registration forms and assist persons with the completion of the forms and transmittal to voter registration authorities. The Help America Vote Act of 2002 (HAVA), enacted 10 years later, required that voting systems used in federal elections meet requirements on accessibility for individuals with disabilities, including nonvisual accessibility for the blind and visually impaired. HAVA provided funds to states to replace problem-plagued punch card and mechanical lever voting machines, generally not accessible to disabled voters, and required jurisdictions to provide at least one direct recording electronic (DRE) voting system or other voting system equipped for individuals with disabilities at each polling place. HAVA also provided funding for improvements to make polling places accessible to individuals with disabilities and to provide disabled voters with the same opportunities for access and participation, including privacy and independence, as for other voters. Survey Results Table 14 presents data on polling place accessibility from question 21 of the Election Day Survey. In the table, the numbers of accessible polling places are calculated as percentages of the total number of polling places. The column headings in Table 14 are as follows: Table 14 Column Headings. Disability Col. Heading Description 1 Code State census code 2 Name Respondent to Election Day Survey 3 Jurisdiction Number of local election jurisdictions from survey question 22 4 Total Number of Precincts Number of precincts from survey question 19 5 Cases Number of jurisdictions that responded to question 19 6 Total Number of Polling Places Number of polling places from survey question 20 7 Cases Number of jurisdictions that responded to question 20

240 2004 Election Day Survey Report, Part 2 Survey Results Disability, Page 14-3 September 27, 2005 Table 14 Column Headings (cont.) Col. Heading Description 8 Number of Polling Places Accessible, Wheelchair Number of polling places accessible by wheelchair from survey question 21a 9 Cases Number of jurisdictions that responded to question 21a 10 Percent of Precincts Accessible, Wheelchair Number of polling places accessible by wheelchair (col. 8) divided by number of precincts (col. 4) 11 Cases Number of jurisdictions that responded to questions 4 and 21a 12 Percent of Polling Places Accessible, Wheelchair Number of polling places accessible by wheelchair (col. 8) divided by number of polling places (col. 6) 13 Cases Number of jurisdictions that responded to questions 6 and 21a 14 Number of Polling Places Accessible, Visually Impaired Number of polling places where the visually impaired can cast a private ballot from survey question 21b 15 Cases Number of jurisdictions that responded to question 21b 16 Percent of Precincts Accessible, Visually Impaired Number of polling places where the visually impaired can cast a private ballot (col. 14) divided by number of precincts (col. 4) 17 Cases Number of jurisdictions that responded to questions 4 and 21b 18 Percent of Polling Places Accessible, Visually Impaired Number of polling places where the visually impaired can cast a private ballot (col. 14) divided by the number of polling places (col. 6) 19 Cases Number of jurisdictions that responded to questions 6 and 21b 20 # of Polling Places Accessible, Physically Disabled Number of polling places with an accessible voting system for physically disabled voters from survey question 21b 21 Cases Number of jurisdictions that responded to question 21b 22 Percent of Precincts Accessible, Physically Disabled Number of polling places with an accessible voting system for physically disabled voters (col. 20) divided by the number of precincts (col. 4) 23 Cases Number of jurisdictions that responded to questions 4 and 21c 24 Percent of Polling Places Accessible, Physically Disabled Number of polling places with an accessible voting system for physically disabled voters (col. 20) divided by the number of polling places (col. 6) 25 Cases Number of jurisdictions that responded to questions 6 and 21c

241 2004 Election Day Survey Report, Part 2 Survey Results Disability, Page 14-4 September 27, 2005 Analysis of Survey Results The following is our analysis of the data in Table 14 for each of the 18 cross-tabulation factors described earlier in this report. A description of each factor follows a general summary and a statelevel summary of the survey data. 1) Regions 10) Changed Voting Equipment since ) Urban to Rural 11) Statewide Voter Registration Database 3) Size of Jurisdiction 12) Election Day Registration 4) Race and Ethnicity 13) Provisional Ballot Acceptance 5) Median Income 14) No Excuse Absentee Balloting 6) High School Education 15) Early Voting 7) Section 203 Language Minority Requirements 16) Battleground States 8) Section 5 Preclearance of Voting Procedures 17) Presidential Margin of Victory 9) Type of Voting Equipment 18) Red versus Blue Jurisdictions This analysis is based only on data that was reported to the EAC on the Election Day Survey. Many state responses to a survey question or part of a question did not cover all local election jurisdictions. In Table 14 as well as other tables in this report, a jurisdiction was excluded from a statistical calculation if its response was missing for one or more of the data items (i.e., columns) used in the calculation. A column labeled Cases next to each statistical calculation shows the number of jurisdictions covered by that calculation. Summary Overall, 92.4 percent of the polling places and 71.7 percent of the precincts in this nation were reported to be wheelchair-accessible. However, this information reflects data from only half of the nation s election jurisdictions. Fewer than a quarter of the precincts and polling places were reported by the states as being locations where a visually impaired voter could cast a ballot in private. A physically disabled voter could cast a ballot on an accessible voting system in only about half the precincts and slightly more than 70 percent of the polling places. States Most of the states that reported accessibility information claimed that nearly all polling places were wheelchair-accessible. Twenty-three of the 26 states that responded said that more than 90 percent of their polling places would allow a voter using a wheelchair to cast a ballot. Most of the other states reported that more than 80 percent of their polling locations were accessible. Virginia reported the lowest percentage, but it was most likely because not all the jurisdictions responded to the survey. Regions All four regions of the nation reported that more than 90 percent of polling places were wheelchairaccessible. Of the jurisdictions that reported, those in the West had the highest percentage of accessible polling places, at over 99 percent. The South had the lowest percentage, yet 91.8 percent of polling places were wheelchair-accessible. Urban to Rural Suburban locations reported the highest percentage of accessible polling places, at over 98 percent. Rural jurisdictions reported the lowest accessible rate, but that was only 92.2 percent.

242 2004 Election Day Survey Report, Part 2 Survey Results Disability, Page 14-5 September 27, 2005 Size of Jurisdictions The smallest jurisdictions reported the lowest percentage of accessible polling places about 90 percent of polling locations could accommodate wheelchairs. As jurisdictions got larger in size, their accessibility scores improved. Race and Ethnicity Predominantly African American jurisdictions in this nation reported that only 81 percent of their polling locations were accessible, while predominantly Non-Hispanic White jurisdictions reported 94 percent accessible sites. Predominantly Hispanic jurisdictions reported that nearly all polling places were accessible. Median Income The poorest jurisdictions reported the lowest accessibility of polling places. While accessibility rates rose as income levels increased, they peaked at the upper middle class jurisdictions and then declined slightly in the highest income jurisdictions. High School Education Education levels in jurisdictions did not appear to have an impact on whether polling locations were accessible. Section 203 Language Minority Requirements Surprisingly, jurisdictions that are covered by the language minority provisions of the Voting Rights Act had a much higher rate of accessibility than jurisdictions that were not covered. Section 5 Preclearance of Voting Procedures Whether or not the jurisdiction was covered by Section 5 of the Voting Rights Act correlated to very small differences in accessibility. Type of Voting Equipment The only difference in wheelchair accessibility is in paper ballot jurisdictions, which average only 91 percent accessible, compared to the mid-90s for all other types of voting equipment. For the visually impaired question, only 41 percent of optical scan jurisdictions reported their polling places were accessible. But because this score is so much lower than other types of equipment jurisdictions, it is very possible that the low score reflected the nature of optical scan equipment, which is more difficult to use for visually impaired voters. Changed Voting Equipment since 2000 Jurisdictions that have changed their voting equipment in the past four years report a slightly higher rate of accessibility in their polling locations than those jurisdictions that have not changed equipment. Statewide Voter Registration Database Jurisdictions in states with a statewide voter registration system in place for the 2004 election actually reported a higher rate of accessibility than states that did not have a statewide registration system.

243 2004 Election Day Survey Report, Part 2 Survey Results Disability, Page 14-6 September 27, 2005 Election Day Registration Jurisdictions that had Election Day registration had a slightly higher rate of accessible polling places than nonelection day registration jurisdictions. Provisional Ballot Acceptance Jurisdictions that accept provisional ballots from anywhere in the jurisdiction had a slightly higher rate of accessibility than jurisdictions that required voters to cast provisional ballots in their home precincts. No Excuse Absentee Balloting Jurisdictions that allowed no excuse absentee balloting had a higher rate of accessibility than jurisdictions that did not. Early Voting Jurisdictions that allowed early voting had a slightly higher accessibility rating than jurisdictions that did not allow early voting. Battleground States There was no real difference in accessibility between jurisdictions in battleground or nonbattleground states. Presidential Margin of Victory There was no real difference in accessibility between those jurisdictions that reported different margins of victory than in other jurisdictions. [This statement is vague; what does different margins of victory mean?] Red versus Blue Jurisdictions Jurisdictions that were carried by Senator Kerry in the 2004 presidential election had slightly higher rates of accessibility than jurisdictions won by President Bush. REFERENCES Kimberling, William C Access for All. The FEC Journal of Election Administration 15(Autumn). U.S. Department of Justice. Civil Rights Division. Disability Rights Section ADA Checklist for Polling Places. February.

244 EAC Election Day Survey Disability 2004 General Election Updated: 09/19/ :08:37 Disability Cases = Number of Jurisdictions Reporting Subject Matter Wheel Chair Accessability Visually Impaired Accessibility Physically Disabled Accessability Total Number of Percent of Number of Percent of Number of Percent of Election Total Number of Polling Percent of Polling Polling Percent of Polling Polling Percent of Polling Administration Number of Polling Places Precincts Places Places Precincts Places Places Precincts Places Code Name Jurisdictions Precincts Cases Places Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases 01 Alabama 67 2, , Alaska Arizona 15 2, , , Arkansas 75 2, , , California 58 21, , , , Colorado 64 3, , Connecticut Delaware District of Columbia Florida 67 6, , , , , Georgia 159 3, , Hawaii Idaho Illinois , , , Indiana 92 5, , Iowa 99 1, , , , Kansas 105 3, , , Kentucky 120 3, , Louisiana 64 4, , Maine Maryland 24 1, , , , , Massachusetts 351 2, , , Michigan 83 5, , , , Minnesota 87 4, Mississippi 82 1, , , , Missouri 116 5, , , Montana Nebraska 93 1, , , Nevada 17 1, New Hampshire New Jersey 21 6, , , , New Mexico New York 58 15, , , North Carolina 100 2, , , , , North Dakota Ohio 88 11, , , Oklahoma 77 2, , , Oregon 36 1, Pennsylvania Rhode Island South Carolina 46 2, South Dakota Tennessee 95 2, , , , Texas 254 8, , , , , Utah 29 1, , , Vermont Virginia 134 2, , , Washington 39 6, , West Virginia 55 1, Wisconsin 1,910 3,563 1,253 2,686 1,596 2,093 1, , , Wyoming American Samoa 1 66 Guam 1 72 Puerto Rico 110 1, , , , , Virgin Islands Total 6, ,252 5, ,754 5,180 62,670 3, , ,563 15, ,915 1, , ,206 Maximum 1,910 21,857 1,253 14,467 1,596 8,639 1, , ,240 2, , Average 119 3, , , , Minimum Election Data Services, Inc. Confidential 9/22/05 Page 1 of 4

245 EAC Election Day Survey Disability 2004 General Election Updated: 09/19/ :08:37 Disability Cases = Number of Jurisdictions Reporting Subject Matter Wheel Chair Accessability Visually Impaired Accessibility Physically Disabled Accessability Total Number of Percent of Number of Percent of Number of Percent of Election Total Number of Polling Percent of Polling Polling Percent of Polling Polling Percent of Polling Administration Number of Polling Places Precincts Places Places Precincts Places Places Precincts Places Code Name Jurisdictions Precincts Cases Places Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Election Administration Voting Equipment Used in 2004 General Election None / Unknown , , , , , Punch card , , , , Lever , , , , Paper 1,734 5,704 1,573 3,416 1,183 2, Optical scan 2,541 69,370 2,405 46,265 2,185 31,582 1, , ,813 3, , Electronic , , , , , Multiple Systems , , , , , Changed Voting Equipment Since 2000 General Election Yes 1,753 46,241 1,296 31,649 1,269 17, , , No 4, ,011 4,100 82,105 3,911 44,987 2, , ,753 8, ,702 1, ,003 State Wide Voter Registration System in Place Yes 1,335 33,575 1,153 20,815 1,133 10, , No 5, ,677 4,243 92,939 4,047 51,831 2, , ,799 15, , Election Day Registration Yes 2,823 9,704 1,924 3,794 1,663 3,205 1, , , No 3, ,548 3, ,960 3,517 59,465 2, , ,256 15, ,939 1, ,148 Provisional Ballot Acceptance In Overall Jurisdiction 1,162 65,986 1,080 44,212 1,070 21, , , In Precinct Only 4, ,295 3,504 66,513 3,902 38,638 2, , ,981 9, , None 1,056 7, , , , , No Excuse Absentee Balloting Yes 3,781 70,535 3,106 47,225 2,922 25,867 1, , ,993 6, , No 2, ,717 2,290 66,529 2,258 36,803 1, , ,570 9, , Early Voting Allowed Yes 1,701 69,882 1,683 51,609 1,618 30, , , No 4, ,370 3,713 62,145 3,562 31,819 2, , ,588 5, , Covered By Section 203, Language Minority Requirements Yes , , , , , No 6, ,201 4,953 77,656 4,737 42,187 3, , ,224 9, ,265 1, ,112 Covered By Section 5 of Voting Rights Act Yes , , , , , No 5, ,276 4,541 88,074 4,377 47,114 3, , ,151 9, , Election Data Services, Inc. Confidential 9/22/05 Page 2 of 4

246 EAC Election Day Survey Disability 2004 General Election Updated: 09/19/ :08:37 Disability Cases = Number of Jurisdictions Reporting Subject Matter Wheel Chair Accessability Visually Impaired Accessibility Physically Disabled Accessability Total Number of Percent of Number of Percent of Number of Percent of Election Total Number of Polling Percent of Polling Polling Percent of Polling Polling Percent of Polling Administration Number of Polling Places Precincts Places Places Precincts Places Places Precincts Places Code Name Jurisdictions Precincts Cases Places Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Demographics Region Northeast 1,710 25,068 1,230 13, , , , South 1,423 48,810 1,408 37,805 1,302 23, , , Midwest 2,902 55,993 2,243 35,954 2,490 23,186 1, , , , West , , , , Territories 113 1, , , , , Urban to Rural Urban , , , , , Suburban , , , , , Small Towns 1,710 41,994 1,421 28,085 1,283 12, , , Rural 3,307 32,252 2,780 21,938 2,548 13,539 1, , ,753 1, , Not Available - Territories 113 1, , , , , Size of Jurisdiction (VAP) < 1,000 1,761 2,118 1,229 1,350 1, >=1,000 to <3,500 1,165 2, , , >=3,500 to <10,000 1,043 8, , , , >=10,000 to <50,000 1,704 35,443 1,554 25,830 1,508 14, , , >=50,000 to <250, , , , , , >=250,000 to <1,000, , , , , , >=1,000, , , , , , Not Available 144 1, , , , , Race and Ethnicity Predominently NH White 6, ,698 5, ,108 4,925 54,823 3, , ,370 12, ,467 1, ,053 Predominently NH Black 85 2, , Predominently NH Native American Predominently Hispanic 50 7, , , , , Not Available 145 1, , , , , Median Income < $25, , , , >=$25,000 to <$30, , , , , , >=$30,000 to <$35,000 1,372 23,424 1,197 16,639 1,076 9, , , >=$35,000 to <$40,000 1,215 40,250 1,056 24, , , , >=$40,000 to <$45, , , , , , >=$45,000 to <$50, , , , , , >=$50,000 1,180 36, , , , , Not Available 151 1, , , , , High School Education < 60% 126 2, , >=60% to <70% , , , , , >=70% to <80% 1,646 51,393 1,411 32,782 1,319 18, , , >=80% to <90% 3,111 87,644 2,502 56,581 2,410 29,352 1, , ,711 6, , >=90% , , , , Not Available 151 1, , , , , Election Data Services, Inc. Confidential 9/22/05 Page 3 of 4

247 EAC Election Day Survey Disability 2004 General Election Updated: 09/19/ :08:37 Disability Cases = Number of Jurisdictions Reporting Subject Matter Wheel Chair Accessability Visually Impaired Accessibility Physically Disabled Accessability Total Number of Percent of Number of Percent of Number of Percent of Election Total Number of Polling Percent of Polling Polling Percent of Polling Polling Percent of Polling Administration Number of Polling Places Precincts Places Places Precincts Places Places Precincts Places Code Name Jurisdictions Precincts Cases Places Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Accessible Cases Political Battleground States in 2004 Presidential Election Yes 3,093 59,123 2,113 33,037 2,309 19,833 1, , ,720 3, , No 3, ,129 3,283 80,717 2,871 42,837 1, , ,843 12, , Margin of Victory in 2004 Presidential Election < 2.5% , , , , , >=2.5% to < 5.0% , , , , >=5.0% to < 7.5% , , , , , >=7.5% to < 10.0 % 429 9, , , , >=10.0 % 4, ,787 3,788 83,067 3,664 47,210 2, , ,458 9, , Red vs Blue Jurisdictions Won By in 2004 Presidential Election Bush > 55% 3,115 68,994 2,690 49,173 2,617 27,680 1, , ,740 5, , Bush 50% to 55% , , , , , Bush < 50% 136 1, , Kerry < 50% 150 4, , , , Kerry 50% to 55% , , , , , Kerry > 55% 1,161 49, , , , , Tied Election Data Services, Inc. Confidential 9/22/05 Page 4 of 4

248 PART 3 RECOMMENDATIONS ON FUTURE DATA COLLECTION As we processed and analyzed the survey results, we developed a number of recommendations on the collection of data on future Election Day Surveys. These consist of general recommendations as well as recommendations pertaining to specific tables and categories of election information. General Recommendations 1. Survey timeline. The late distribution of the 2004 Election Day Survey resulted from delays in the creation of the U.S. Election Assistance Commission, as well as the time necessary to get the survey instrument accepted through the Paperwork Reduction Act requirements. We recommend that the EAC distribute the final 2006 Election Data Survey no later than eight weeks before the November election to allow enough time for state election directors to make data requests of local election administrators and for local administrations to set up the systems needed to collect the requested data. Late responses by state election directors were also a problem as some surveys were received more than three months after the deadline. We also recommend that the EAC make sure that deadlines are reasonable and then take steps to encourage state election directors to submit responses by those deadlines. 2. Survey format. Although 2004 Election Day Survey was distributed as a Microsoft Excel spreadsheet, states were free to adjust the format of the spreadsheet as well as to submit documents in other formats. The variety of formats that were used resulted in a very costly and time consuming effort to standardize the survey responses for analysis. We recommend that the EAC incorporate into the design of the 2006 Election Day Survey fill-able cells and other techniques to obtain election data in a more uniform format to reduce the amount of data processing. A more uniform format would also provide greater assurances that the survey responses were tabulated accurately. A password-protected Internet survey could be designed to reduce the amount of processing required for survey responses. By creating an internet-based survey, we could produce quality assurance reports in real time, which will help state election directors identify data entry errors as the data was submitted. 3. Statistics in elections. Back in 1978, the Principal Investigator this project was a subcontractor on a project for the Office of Election Administration of the Federal Election Commission (FEC) devoted to studying the use of statistics in elections.

249 2004 Election Day Survey Report, Part 3 Recommendations, page 2 September 27, 2005 For some reason the results of that study were never published. We recommend that the EAC revive the intent behind the earlier study and undertake a research effort to document the importance of statistics and audits in the elections process. This study could form the basis of that future work. 4. Uniform election information. It is clear that different states and localities keep track of different statistics that use different words to describe them. In other states, the excuse is given that their state law doesn t require the information so they don t collect it. This study has clearly pointed out that there needs to be some uniformity in information on the elections process. We recommend that the EAC use whatever influence it has to ensure that uniformity. To further the goal of having uniform information on the elections process, we also recommend that the EAC convene a meeting of all state election directors to discuss the results of this report. 5. Precinct-level data. Despite the difficulties in gathering the county and jurisdiction level data for this report, the gross level of the data made meaningful analysis of some of the information impossible. Demographic analysis was impaired due to the lack of large enough concentrations of different characteristics. Studying the malfunctions of different voting equipment and their impacts on different groups is impossible when data is at the jurisdiction level. Therefore, we recommend that the EAC undertake the creation of a uniform precinct level database of election information for the country. Collecting basic registration, turnout and election returns will actually cut down in the data being requested via a survey, since much of the basic information is available as the results are being certified. 6. Database files of election results (vote tallying software). One of the difficulties in gathering precinct level information involves the wide variety of formats of printed reports produced by vote tallying software. Keypunching of this information means that each county s information has to be analyzed and potentially formatted separately before a single number is entered. As part of the NIST development of voting equipment standards, we recommend that the EAC ensure that all vote tallying software be required to produce a database file of the election results and basic information outlined in this report. The EAC and NIST should outline the minimum geographic identifiers that should be in the database files, as well as the basic information. The basic information would include, among other items, both overvote and undervote counts for each office and contest on the ballot. 7. Election Day Survey Follow-up Review. Due to the time pressures of getting this initial report produced for the EAC and Congress, data holes and errors still exist in the information that forms the basis of this report. In June 2005 we proposed that the EAC fund a continuation of this project to work with the states to find and correct errors and fill in missing information. The proposal was accepted, and in early July, jurisdiction-level spreadsheets with data from the survey were sent to state election directors for review. Responses were received from 26 states by the July 15, 2005, deadline that was established for the follow-up review. These responses were imported to the survey database to update the final version of the Election Day Survey

250 2004 Election Day Survey Report, Part 3 Recommendations, page 3 September 27, 2005 Report. To improve survey coverage rates, the EAC could extend the follow-up review period to collect spreadsheets from the 29 state non-respondents and make further updates to the Election Day Survey Report. New census population estimates that will be released later this summer could also be incorporated into an updated version of the Election Day Survey Report. 8. Election audits. In the accounting field, audits of data are a regular part of their business. The state of New Mexico has an auditing team that compiles and certifies all of their official election results. We recommend that the EAC produce a document or pamphlet that outlines various steps that state and local governments could undertake to ensure they have the proper data for each election. This auditing process should start at the precinct level on Election Day, and flow all the way to the state. Summary of Recommendations in Part 2 The following recommendations were presented in part 2 of the report and are summarized below: 9. Population data. Recommendations on population data in Chapter 1 were as follows: 9.1. That the EAC request information on voter eligibility requirements in the states and any changes to state law since the last federal election to better measure the eligible population That the EAC request jurisdictions to provide estimates of eligible and ineligible persons. For example, some states use lists of felons to purge their registration rolls. Some states may have information on the number of overseas eligible citizens from sources such as tax records That the EAC work with U.S. Census Bureau to obtain population and voting age population estimates and projections that will correspond with the general election calendar for counties and townships in Michigan, Wisconsin, and the six New England states. This will provide a uniform base from which an election analysis can be preformed. In past years, the Census Bureau produced state-level projections of voting age populations prior to the November general elections. This dataset was dropped in The EAC should encourage Congress and the Census Bureau to re-instate the program so that state and local governments would have benchmark data by which to compare their own information Because the territories of Guam, American Samoa and the Virgin Islands are now covered by the Help America Vote Act (HAVA), the Census Bureau should be encouraged to include those jurisdictions in their population estimates program so that post-decennial census population and voting age population data would be available.

251 2004 Election Day Survey Report, Part 3 Recommendations, page 4 September 27, Voter registration. Recommendations on voter registration data in Chapter 2 were as follows: That the EAC ask states and local election jurisdictions to keep counts of both active and inactive registrants and to report both numbers, plus the total number of registered voters to the EAC. The differences in how states and localities report registration counts has a significant impact on any study of voting. Whether a jurisdiction uses only active voters or chooses to combine inactive and active voters has a huge bearing on how the jurisdiction is perceived to carry out it s registration responsibilities. When compared to voting age population estimates, registration rates can vary wildly dependent upon how a state reports its registration numbers That the EAC also collect the number of persons who registered to vote on Election Day for those jurisdictions in states with Election Day registration That the EAC create a table of the eligibility requirements for both voter registration and for voting in each state. These requirements would produce variables for further analysis That the EAC investigate the rules and procedures used in each state under which a registered voter is moved from active to inactive status. Data on the number of voters who are removed from a voter registration file, as well as the number of voters that were transferred to another jurisdiction could also be collected. 11. Voter turnout. Recommendations on voter turnout data in Chapters 3 and 4 were as follows: That the EAC collect information on the different deadlines used by states for close of registration, and use that information to further investigate their impact on turn-out rates That the EAC use its influence to get all states and local governments to compile a true voter turnout number for each election. Despite the EAC's efforts over the past year, the American people still doesn't know the total number of persons who showed up and participated in the 2004 election. A handful of states and local jurisdictions still don't collect an actual turnout number, instead believing the incorrect assumption that one just needs to tally up the number of votes received by all candidates for the highest office. This study, once again, points out the fallacy of such a belief That the EAC on future surveys make it clear to states and local jurisdictions when component questions are part of the whole election process and should sum to 100 percent. Clearer instructions and more timely informa-

252 2004 Election Day Survey Report, Part 3 Recommendations, page 5 September 27, 2005 tion should be conveyed to the states and jurisdictions so that counts on the various methods of voting can be kept separately That the EAC consider merging the UOCAVA (Military and Overseas Absentee Ballot) survey into the Election Day survey That the EAC should collect actual election results for all offices at the precinct level to facilitate determining the accuracy of data compiled via survey instruments. This tends to be a standard report released by all jurisdictions in the country, and therefore, is available on a quick turn-around once the returns have been certified. 12. Absentee ballots. Recommendations on absentee ballots in Chapter 5 were as follows: That the EAC combine UOCAVA questionnaire with the Election Day Survey. We believe that most jurisdictions disregarded the Election Day Survey instructions to separate military and overseas absentee statistics from other absentee statistics and provided statistics on all absentee voters. Combining the two surveys would reduce confusion That the EAC collect additional information on how absentee ballots are requested, returned and counted. For example, some states allow permanent absentee balloting, which may be related to increased rates of absentee ballot requests. Some states permit voters with absentees to return absentee ballots to the polling place on Election Day, and in some cases these ballots may not have been counted as a returned absentee ballot That the EAC clarify the absentee ballot definition, particularly for the case of Oregon, which runs its elections by mail That the EAC ask all states keep counts of absentee returns separate from results cast at the polling place. However, absentee results need to be available at the smallest geographic level (preferable by precinct), so that full demographic analysis can take place. Reporting absentee returns at only the county level or at a ballot style level muddles any meaningful analysis possible. 13. Provisional ballots (Chapter 6). Recommendations on provisional ballots were as follows: That the EAC collect separate statistics on challenged and provisional ballots That the EAC collect state and jurisdictional rules and statistics regarding how ballots cast by first time voters without identification are processed on Election Day That the EAC collect statistics on the number of first time voters who do not provide identification, how many vote a provisional or challenged ballot, and how many of these types of ballots are counted

253 2004 Election Day Survey Report, Part 3 Recommendations, page 6 September 27, That the EAC collect information on the procedures used to ascertain the validity of a provisional ballot for example, what is the timing of the verification process That the EAC collect information regarding how jurisdictions notify individuals concerning the disposition of their provisional ballot. 14. Drop-off, overvotes, and undervotes. Recommendations on drop-off and overvotes and undervotes in Chapters 7 and 8 were as follows: That the EAC gather actual election results to better understand how patterns of electoral competition factor into drop-off. Many who study and opine about civic engagement in the United States focus on participation in elections as the most fundamental act of civic engagement. Yet little data has been collected regarding voters overall voting experience and the decisions they face as they work their way down the ballot That the EAC establish a clearinghouse for the collection of sample ballots or images of actual ballot images used within jurisdictions. Documenting and observing the appearance of actual ballots, coupled with actual election results and turnout data, will aid understanding of the various factors that ultimately lead to voter choices to participate in elections appearing on a ballot That the EAC encourage all state and local election officials to produce separate counts of overvotes and undervotes, for at least their own internal review of the election. A number of jurisdictions did not provide overvotes and undervotes, and a full-scale audit of election results should incorporate the study of both all overvotes and all undervotes. Too many times vendors have told election officials that providing such information only confuses the process. But the data help form the heart of determining whether or not problems exist in a given election. 15. Voting equipment. Recommendations on voting equipment in Chapters 9, 10 and 11 were as follows: That the EAC collect information on who provides on-going support of the voting system to the jurisdictions. In many instances, this will be the same as the manufacturer of who sold them the equipment. This may assist in filling out the blank information received by the EAC for half the nation on who is the manufacturer That the EAC should seek more detailed information on voting equipment devices and manufacturers from all jurisdictions. The use of generic voting equipment type categories by some states prevents a complete picture of the voting equipment market in the United States.

254 2004 Election Day Survey Report, Part 3 Recommendations, page 7 September 27, 2005 The growing use of multiple voting systems in the same jurisdiction has the potential to prevent a meaningful analysis of future voting, such as has been conducted in this report, unless jurisdictions keep election returns, over and under votes, and turn-out information separate for each of the different types of voting equipment in use. This has the potential of requiring jurisdictions to keep different tallies for each precinct for the machines in use. The EAC should investigate this growing trend and work with state and local election officials to arrive at a satisfactory solution to the problem That the EAC collect information on where ballots are tallied, be they at the precinct or at a central county location. This would allow a more complete analysis to be done on whether drop-off or overvotes are less likely to occur when the voters are present and have an opportunity to correct voting mistakes That the EAC collect information on the number of polling booths or actual voting devices that are used at election time. Confusion over question wording in the 2004 survey prevents proper analysis from being conducted on one potential cause of the long lines in various states That the EAC institute a more extensive program designed to investigate reported voting equipment problems. During the late 1970s, the National Institute of Standards and Technology (NIST) sent investigator Roy Saltman to a number of jurisdictions around the country that had problems with punch card voting systems. His very detailed reports provided background information, an unbiased description of the problems encountered, reasons on why the problems occurred, and descriptions of solutions instituted. With the wide ranging rumors and reports of voting equipment problems that came out of the 2004 elections, there is a lack of full information to substantiate or dispel the rumors. 16. Poll workers. Recommendations on poll workers in Chapter 12 were as follows: That the EAC change how the number of poll workers is collected. States vary with regards to requiring poll workers to work all day or in shifts. This administrative procedure by itself would be valuable information to collect. However, in calculating number of poll workers per polling place or precinct, a comparable metric needs to be formulated across jurisdictions, such as the average number within polling places during the day. Similarly, the number of polling places or precincts with staffing concerns should be considered in terms of inadequate coverage of a shift or for the entire day That the EAC collect information about poll worker training and special skills required of poll workers, such as: (a) How are poll workers trained? Is training mandatory? And how many hours is a typical training class? (b) Is

255 2004 Election Day Survey Report, Part 3 Recommendations, page 8 September 27, 2005 multilingual training provided for poll workers in Section 203 covered jurisdictions? And (c) are they compensated for their time and at what rate? 17. Polling places. Recommendations on polling places in Chapter 13 were as follows: That the EAC collect the number of early voting and Election Day polling places. With the increasing popularity of early voting, the distinction between early voting and Election Day precincts will need to be carefully defined and the growth or decline of polling places will need to be monitored That the EAC begin asking about the existence of vote centers, how many precincts they cover, and determine whether returns and other data are consolidated in how they are reported or are able to be kept separate for the individual precincts. The advent in the past year of the concept of vote centers, or locations where voters can come from multiple precincts, is a new development in election administration and something that merits study That the EAC collect information on criteria for establishing precincts. For example, what does each state law require as a maximum or minimum size for a voting precinct? This information would be useful to identify standards and best practices among jurisdictions That the EAC collect information on the number of consolidated polling places, i.e., polling places servicing more than one voting precinct, and collect procedures for the establishment of consolidated polling places. Consolidated polling places were identified as one potential cause of voter confusion that might lead to the casting of an invalid provisional ballot That the EAC collect information on individual voting precincts and polling places, e.g., the number of registered voters and the number of votes cast in each voting precinct and polling place. A populous jurisdiction may have several hundred voting precincts within its boundaries, and jurisdiction averages may mask significant variation across voting precincts and polling places. If the unit of analysis was the precinct or polling place, additional characteristic may be collected, such as ease of access of polling places, recent changes to precincts boundaries, and polling place location That the EAC collect information on split precincts in the states that use them. Split precincts or polling places were identified as one potential cause of voter confusion that might lead to the casting of an invalid provisional ballot.

256 2004 Election Day Survey Report, Part 3 Recommendations, page 9 September 27, Disability: Recommendations on polling place accessibility in Chapter 14 were as follows: That the EAC clarify the wording of questions about accessible polling locations so that it is clear the information being sought relates to the physical polling site and not the type of equipment used.

257 Appendix A Survey Instrument

258 UNITED STATES ELECTION ASSISTANCE COMMISSION ELECTION DAY DATA SURVEY The following form and its attachments should be completed on-line or in electronic format. Completed forms should be returned to the United States Election Assistance Commission via to on or before the sixtieth day following the federal election. Please complete all of the fields below. Specific instructions relative to certain fields are found at the end of this form. If your question or concern is not answered in the instructions section of this form, please contact Brian Hancock at RESPONDANT INFORMATION: Name of the responding State: Date response is submitted: Name of responding official: Title of responding official: Address of responding official: Address: Phone Number: Fax Number: VOTER REGISTRATION: Please respond to the following questions on the attached spreadsheet or in similar format that can be imported to a spreadsheet or database file such as a comma separated values (.csv) format. 1a. Total number of registered voters (active) by county/local election jurisdiction 1b. Total number of registered voters (inactive) by county/local election jurisdiction ELECTION RESULTS: Please respond to the following questions on the attached spreadsheet or in similar format that can be imported to a spreadsheet or database file such as a comma separated values (.csv) format. 2a. Total number of ballots counted statewide 3a. Total number of ballots cast in polling places (state-wide) on election day (for Oregon by mail) 4a. Total number of requested absentee ballots 5a. Total number of absentee ballots returned 2b. Total number of ballots counted by county/local election jurisdiction 3b. Total number of ballots cast on election day by county/local election jurisdiction 4b. Total number of requested absentee ballots by county/local election jurisdiction 5b. Total number of absentee ballots returned by county/local election jurisdiction

259 6a. Total number of absentee ballots counted 6b. Total number of absentee ballots counted by county/local election jurisdiction 6c. Total number of absentee ballots that 6d. Identify the five most common reasons were not counted that absentee ballots were rejected 7a. Does your state conduct early voting? 7b. Total number of early voting ballots counted 7c. Total number of early voting ballots counted by county/local election jurisdiction 8a. Total number of provisional ballots cast 8b. Total number of provisional ballots cast by county/local election jurisdiction 9a. Total number of provisional ballots 9b. Total number of provisional ballots counted counted by county/local election jurisdiction 9c. Identify the five most common reasons that provisional ballots were rejected 10. Total number of undervotes reported in 11. Total number of overvotes reported in each federal contest by county/local election each federal contest by county/local election jurisdiction jurisdiction 12. Total number of votes cast for all candidates in each federal contest by county/local election jurisdiction VOTING EQUIPMENT: Please respond to the following questions by attaching a file in any of the following formats:.doc,.txt, or.csv. 13. Provide a listing of the types of voting equipment in use in each county of the State including the type of voting system, manufacturer, number of units used in each county/local election jurisdiction, the software version (if applicable), and an indication as to whether the voting system has or has not previously been used in a Federal election in that jurisdiction. 14. Identify by county and precinct, if available, where any of the following voting machine malfunctions occurred. Please identify if the voting machine was returned to service in the November 2, 2004 election. 14a. Power failure 14b. Broken counter 14c. Computer failure 14d. Printer failure 14e. Screen failure 14f. Fatal damage to machine 14g. Modem failure 14h. Scanner failure 14i. Ballot encoder/activator failure 14j. Audio ballot failure 14k. Other (please specify) POLL WORKERS: Please respond to the following questions on the attached spreadsheet or in a similar format that can be imported to a spreadsheet or database file such as a comma separated values (.csv) format. 15a. Total number of poll workers who served in the State on November 2, b. Total number of poll workers who served in each county/local election jurisdiction on November 2, What is the required number of poll workers per precinct/polling place as established by law or regulation? 17a. By county/local election jurisdiction, how many precincts/polling places did not have the required number of poll workers? 17b. In any county/local election jurisdiction where a deficit of poll workers existed, identify the number of additional poll workers needed to meet the requirement.

260 VOTING JURISDICTIONS: Please respond to the following questions on the attached spreadsheet or in a similar format that can be imported to a spreadsheet or database file such as a comma separated values (.csv) format. 18. Identify what constitutes a local election jurisdiction in your State (e.g. county, parish, township, city). 19. Total number of precincts by county/local election jurisdiction 20. Total number of polling places by county/local election jurisdiction 21a. Total number of polling places by county/local election jurisdiction that can be accessed by a voter who uses a wheelchair 21b. Total number of polling places by county/local election jurisdiction where a visually impaired voter can cast a private ballot 21c. Total number of polling places where a physically disabled voter can cast a ballot on an accessible voting system. SOURCES OF INFORMATION: Please respond to the following questions by attaching a file in any of the following formats:.doc,.txt, or.csv. 22. Total number of local election jurisdictions that provided information for purposes of responding to this survey 23. Provide the name and contact information for each local election jurisdiction official that provided information for purposes of responding to this survey. 24. Identify any other sources of information used to respond to this survey other than those provided in response to questions 22 and 23. Instructions: Please answer every question. Do not leave any questions blank. The appropriate answer may be 0, none, or N/A. This survey seeks information on both a State and local election jurisdiction level. A spreadsheet has been attached for your convenience in responding to the majority of the questions, above. However, States may provide the same information in a similar format through any.csv formatted file. Please add additional columns where necessary to report additional Congressional or Senatorial district information and to accommodate all counties/local election jurisdictions in the State. Definitions: The following are specific instructions and definitions for your use in completing the numbered questions in the form, above: 1. Provide by county/local jurisdiction, only, the number of registered voters. If your state differentiates between active and inactive voters, place each number in the respective column on the attached spreadsheet. If your state does not differentiate, place results in the active column.

261 2. The number provided in response to this question should include all ballots that were counted during election day, absentee, early voting or late counting for the November 2, 2004 election (e.g., paper, electronic, military, absentee, and provisional ballots) 3. The number provided in response to this question should include all ballots cast and counted during election day voting (at the polls). This number does not include the number of absentee or early voting ballots counted. 4a. 6d. Absentee voting is defined as voting prior to election day which requires that the voter meet qualifications other than those generally required to register to vote. The numbers provided in response to questions 4a. 6b. should not include ballots requested by military and overseas voters. The number should reflect only those non-military and overseas absentee ballots that were requested, returned, cast and counted, respectively. 6e. Identify the most common reasons for rejecting an absentee ballot. The response to this question can be provided in any electronic format. 7. Early voting is defined as any voting that occurred prior to November 2, 2004 for which there were no eligibility requirements. For example, the voter did not have to attest that he/she would be absent from the voting jurisdiction on the day of the election. 8. The number provided in response to this question should include the total number of ballots cast in the State s program for contingent or provisional ballots that complies with section 302(a) of the Help America Vote Act. 9a. The number provided in response to this question should include the total number of ballots identified in response to question 8 that were verified as having been cast by eligible voters and were counted in the November 2, 2004 election. 9b. The number provided in response to this question should include the total number of ballots identified in response to question 8 that were not verified as having been cast by eligible voters and which were not counted in the November 2, 2004 election. 9c. Identify the most common reasons for rejecting a provisional ballot. The response to this question can be provided in any electronic format. 10. An undervote occurs at any time when a voter makes less than that allowed number of selections in a single race/contest or when a voter votes on less than all of the races/contests for which he/she is eligible to vote. 11. An overvote occurs when a voter makes more than the permitted number of selections in a single race/contest or when a voter makes a selection in a race/contest on which he/she was not eligible to vote. 12. Report all votes cast for all candidates in the presidential, senatorial and congressional contests, respectively. If response to this question is made using the sample spreadsheet, columns must be added so that each senatorial and congressional contest is reported separately. 13. Respond to question 13 by identifying in an electronic document (.doc,.txt. or.csv format) the county and/or precinct, the type of voting system (i.e., punch card, lever, optical scan or direct record electronic (DRE)), the manufacturer of the equipment used, and the number of units in use in the county (and/or precinct, if available), the software version (if applicable) and an indication as to whether this technology was used in a Federal election in that jurisdiction prior to November 2, The answer to this question should identify the location (county or precinct, if available) and number of occurrences of each type of machine malfunction that occurred on November 2, 2004 or during any absentee or early voting period for the November 2, 2004 election. 14a. Power failure means any interruption or failure of the power system of the voting system that would render the voting machine incapable of counting votes for more than 5 minutes during election day, absentee or early voting. 14b. Broken counter means with reference to a lever voting system the malfunction of the counting mechanism that renders the voting system incapable of counting additional votes on any votable position on the machine.

262 14c. Computer failure means any malfunction, disablement or interruption of the software, hardware or firmware that makes up the voting unit such that the unit is incapable of presenting the ballot, recording votes or printing and/or tabulating results. 14d. Printer failure means any malfunction or interruption of the printer hardware, software or mechanical components that constitute the mechanism for creating a printed result of all races voted on a single or on multiple voting machines. Printer failures shall include printers on both electronic and mechanical or lever voting machines. 14e. Screen failure means with reference to a DRE a malfunction or interruption of the screen display or indicator lights such that the DRE cannot accurately indicate to the voter which choices have been made or which races on which the voter is eligible to vote. 14f. Fatal damage to a machine means the damage or destruction of a voting machine that renders it incapable of functioning to record votes or print results of voting. 14g. Modem failure means the malfunction or interruption of modem or the computer hardware or software using the modem to transmit results to a central counting location such that the modem is rendered incapable of transmitting results. 14h. Scanner failure is the malfunction or interruption of a paper ballot reading device that renders it incapable of counting votes or renders the result tabulated by the reader inaccurate. 14i. Ballot encoder/activator failure with reference to a DRE means the malfunction or interruption of that piece of electronic equipment that encodes a smart card or other similar device with the voter s ballot or critical demographic data that allows the voting system to access the proper ballot for the individual voter. 14j. Audio ballot failure with reference to a DRE means any malfunction or interruption of the hardware, software or peripherals that renders the voting machine incapable of playing an audio version of the ballot. 14k. Other refers to any voting machine malfunction that does not fall within the categories established in 14a 14i. 15. The answer to this question should include the number of persons who served in all polling places in the State as poll workers, election judges, wardens, commissioners or other similar term that refers to the person or persons who verify the identity of a voter; assist the voter with signing the register, affidavits or other documents required to cast a ballot; assist the voter by providing the voter with a ballot or setting up the voting machine for the voter; and serving other functions as dictated by state law. The answer to this question shall include the head poll worker for each precinct. The response to this question shall not include observers stationed at the polling place. 16. In responding to this question, please provide any prescribed minimum number of poll workers needed to serve in a precinct/polling place on election day. 17. In response to this question, the State shall identify the county and precinct, if available, where less than a full complement of poll workers was present on election day and the number of poll workers that it was short by county or local election jurisdiction. 18. Identify in any electronic format what constitutes a local election jurisdiction in your State (e.g., county, parish, township, city) 19. Precinct is that geographic area to which voters are assigned. 20. Polling place is that physical structure where residents of a precinct go to cast their votes on election day. A polling place includes any structure that houses one or more precincts. 21a. Identify the total number of polling places that are accessible to persons using wheelchairs. 21b. Identify the total number of polling places where voting equipment is used such that a visually disabled voter can cast a private ballot (e.g., a DRE with audio ballot capability or paper ballots printed in Braille). 21c. Identify the total number of polling places where voting equipment is used that is accessible to a physically disabled voter (e.g., a touch screen DRE which can be handed to the

263 voter, a voting machine which can be lowered to allow access to voter using a wheelchair, other paper ballots or voting systems that are accessible to voters with physical disabilities). 23. In response to this question, the State shall identify name, address, phone number and address (if available) of the local election officials or jurisdictions responsible for conducting elections in a specified geographic area that have provided data to assist the State in responding to this survey. 24. All other sources of data shall include information obtained from a state-wide voter registration database or any other public or non-public source.

264 Appendix B Election Glossary Here are definitions or descriptions of election terms, abbreviations, and acronyms in the 2004 Election Day Survey Report: Absentee Ballot Active Voter Ballots Cast Ballots Counted Central Count CVAP Datavote Deadwood DRE Drop-Off Early Voting EAC EDR FVAP General Election Ballot requested by application of a registered voter who expects to be absent from the polls on Election Day Not inactive (see Inactive Voter) Number of ballots cast and processed at a precinct, including an absentee precinct, and not necessarily equal to the number of ballots counted. Number of ballots counted for a precinct at the close of the polls. Processing or counting of ballots on automatic tabulating equipment at a single location, and usually in reference to punch card and optically scanned (marksense) ballots. Citizen Voting Age Population. Persons in an election jurisdiction who are age 18 or older and who are U.S. citizens. Datavote ballot card. A punch card ballot that is printed with a candidate name or answer to a ballot question at each voting position. A Datavote ballot card is inserted into a frame with a movable device for punching out chads at voting positions. (See Votomatic.) Duplicate names, erroneous or obsolete address information, and names of deceased and ineligible persons listed as active voters on voter registration rolls. Direct Recording Electronic. A voting system that records votes by means of a ballot display provided with mechanical or electro-optical components actuated by the voter and where voting data is stored in a removable memory component. In this report, DRE is referred to as an electronic voting system. The difference between the number of ballots counted and the total number of votes for all candidates in a specified contest. Drop-off is a combination of undervotes and overvotes, and is also referred to as the residual vote. Ballot cast by a voter at a designated polling site prior to Election Day. Also referred to as early in-person voting or on-site absentee voting. U.S. Election Assistance Commission Election Day Registration Federal Voting Assistance Program, Office of the U.S. Secretary of Defense An election in which voters, regardless of party affiliation, select candidates for public office or vote on ballot issues.